# Category: Video

As you probably already know we have been wondering how to best handle our documentation and how to provide information to new Crazyflie starters as easy as possible, as you can read in our blogpost of two weeks ago. In the mean time, we also had a chance to think about the results of a poll we had when we discussed about new ways on how to meet our users. We had about 30 responses, but it became clear that many of you are in the need of getting some more knowledge about working with the Crazyflie. The majority voted for online tutorials, and although it might be difficult to do those during the summer holidays, we already started to make step-by-step guides of various parts of the Crazyflie Eco-system.

## Python Library Tutorials

Currently we have started with step-by-step tutorials of the CFLIB (the python library of the crazyflie). Usually we refer to the example pages of the CFLIB, however we feel that many users copy paste parts of these scripts for their own purposes, without understanding what is actually going on. Therefore, in order to move Crazyflie beginners to the starting developer phase, we have made these guides in order to teach exactly what is going on in each module, step-by-step.

These tutorials can be found in the python library documentation. The first tutorial focuses on connecting, logging and parameters, which guides you through the process of connecting the Crazyflie through a python script, starting up logging configurations in two ways (asynchronous and synchronous) and how to read and set parameters.

The second tutorial is about the motion commander and is a logical continuation of the first. A nice thing is that we also show how to build in some protection in your script as well. If the Flowdeck is not attached, you can check that and prevent the Crazyflie from taking off altogether if it does not detect the Flowdeck. This will be a life saver in your future endeavors, and trust me I know from experience ;). Afterwards it will go into how to take off – fly forward and go back to the initial position. The application of the end of the motion commander tutorial, is where we also use the logging functionality to get the actual estimated position of the Crazyflie. With this information, we show how to write an application that create a virtual bounding box where the Crazyflie can bounce around in (like the old windows screensaver).

We are planning to finish this step-by-step guide by adding the multi-ranger to the mix, continuing on the bouncing ball example. After that we will probably start some tutorials on how to use the swarming functionality before moving on to the firmware or the client.

## Work in Progress

The tutorials are still work in progress. So let us know on the forum, python library github repository or as a comment on this blogpost if you see anything wrong or if something is not very clear. This will improve the quality further so that other users can benefit as well. Also once the these step-by-step tutorials are finished we can start working on video based tutorials as well.

Remember, it is possible to contribute your own fixes (or tutorials) to our repositories if you want to. It’s an open source project after all ;)

Modular robotics implies in general flexibility and versatility to robots. In theory, you could design a modular robot basically on the way you would want it to be, by simply adding or removing modules from the already existing robot. Changing the robot configuration by adding more individuals, generally increases the system redundancy, meaning that now, there are probably many different ways to achieve a specific goal. From a naive standard point of view, more modules could imply in practice more robustness due to this redundancy. In fact, it does get more robust by the cost of becoming more complex, and probably harder to control. Added to that, other issues may arise when you take into account that your modular robot is flying, and how physical properties and actuation scales as the number of modules grow.

In the GRASP Laboratory at the University of Pennsylvania, one of our focuses is to allow robots to achieve a specific task. In this work, we present ModQuad-DoF, which is a modular flying platform that enlarges the configuration space of modular flying structures based on quadrotors (crazyflies), by applying a new yaw actuation method that relies on the desired roll angles of each flying vehicle. This research project is coordinated by Professor Mark Yim , and led by Bruno Gabrich (PhD candidate).

## Scaling Modular Robots

Scaling modular robots is a very challenging problem that usually limits the benefits of modularity. The sum of the performance metrics (speed, torque, precision etc.) from each module usually does not scale at the same rate as the conglomerate physical properties. In particular, for ModQuad, saturation from individual motors would increase as the structures became larger leading to failure and instability. When conglomerate systems scale up in the number of modules, the moment of inertia of the conglomerate often grows faster than the increase in thrust capability for each module. For example, the increase in the moment of inertia for a fifth module added to four modules in a line can be approximated by the mass of the module times half the distance to the center squared. This quadratic increase gives us the intuition that the required yaw actuation grows faster than the actuation authority.

## Yaw Actuation

An inherit characteristic of quadrotors is to have their yaw controlled by the drag moments from each propeller. For ModQuad as more modules are docked together, a decreased controllability in yaw is noticed as the structure becomes larger. In a line configuration the structure’s inertia grows quadratically with the distance of each module to the structure center of mass. On the other hand the drag moments produced scales linearly with the number of modules.

The new yaw actuation method relies on the fact that each quadrotor is capable to generate an individual roll enabled by our new cage design. By working in coordinated manner, each crazyflie can then generate structure moments from moment arms provided by the propellers given its roll and its distance from the structure’s center of mass.

## Cage Design

The Crazyflie 2.0 is the chosen platform to enable thrust and attitude to the individual modules. The flying vehicle measures 92×92×29 mm and weights 27 g while its battery lasts around 4 minutes for the novel design proposed. In this work the cage performs as pendulum relative to the flying vehicle. The quadrotor is joined to the cage through a one DOF joint. The cages are made of light-weight materials: ABS for the 3-D printed connectors and joints, and carbon fiber for the rods.

Although the flying vehicle does not necessarily share same orientation as the cage, the multiple connected cages do preserve same orientation relative to each other. With the purpose of allowing such behavior, we used Neodymium Iron Boron (NdFeB) magnets as passive actuators to enable rigid cage connections. Docking is only allowed at the back and front face of the modules, and each one of these faces contains four magnets. Those passive actuators have dimensions of 6.35 × 6.35 × 0.79 mm with a bonding force of 1 kg.

## Conclusions

ModQuad-DoF is a flying modular robotic structure whose yaw actuation scales with increased numbers of modules. ModQuad-DoF has a one DOF jointed cage design and a novel control method for the flying structure. Our new yaw actuation method was validated conducting experiments for hovering conditions. We were able to perform two, four and, six modules cooperatively flying in a line with yaw controllability and reduced loss in thrust. In future work we aim to explore the structure controllability with more robots in a line configuration, and exploring different solutions for the desired roll angles. Possibly, with more modules in the structure, only a few would be required to roll in order to maintain a desired structure yaw. Given that, we could explore the control allocation for each  module in a specific structure configuration, and dependent on its desired behavior. Further, structures that are not constrained to a line will also be tested using the basis of the controller proposed in this work.

## Detailed Video Explanation (ICRA 2020)

This work was developed by:

Bruno Gabrich, Guanrui li, and Mark Yim

## Autonomous Robotics at UW Seattle

The Crazyflie 2.1 was the perfect robotics platform for an introduction to autonomous robotics at the University of Washington winter quarter 2020. Our Bio-inspired Robotics graduate course completed a series of Crazyflie projects throughout the 10 weeks that built our skills in:

• Python
• Robot Operating System (ROS)
• assembling custom sensors
• writing new drivers
• designing and testing control algorithms
• trouble shooting and independent learning

The course was offered by UW Mechanical Engineering’s Autonomous Insect Robotics Laboratory, headed by Dr. Sawyer B. Fuller. The course was supported by PhD candidate Melanie Anderson, who has done fantastic research with her Crazyflie-based Smellicopter. The final project was an opportunity to turn a Crazyflie quadcopter into a bio-inspired autonomous robot. Our three person team of UW robotics grad students included Nishant Elkunchwar, Krishna Balasubramanian, and Jessica Noe.

## Light Seeking Run-and-Tumble Algorithm Inspired by Bacterial Chemotaxis

The goal for our team’s Crazyflie was to seek and identify a light source. We chose a run-and-tumble algorithm inspired by bacterial chemotaxis. For a quick explanation of bacterial chemotaxis, please see Andrea Schmidt’s explanation of chemotaxis on Dr. Mehran Kardar’s MIT teaching page. She provides a helpful animation here.

In both bacterial chemotaxis and our run-and-tumble algorithm, there is a body (the bacteria or the robot) that can:

• move under its own power.
• detect the magnitude of something in the environment (e.g. chemical put off by a food source or light intensity).
• determine whether the magnitude is greater or less than it was a short time before.

This method works best if the environment contains a strong gradient from low concentration to high concentration that the bacteria or robot can follow towards a high concentration source.

The details of the run-and-tumble algorithm are shown in a finite state machine diagram below. The simple summary is that the Crazyflie takes off, begins moving forward, and if the light intensity is getting larger it continues to “Run” in the same direction. If the light intensity is getting smaller, it will “Tumble” to a random direction. Additional layers of decision making are included to determine if the Crazyflie must “Avoid Obstacle”, or if the source has been reached and the Crazyflie quadcopter should “Stop”.

## Crazyflie Hardware

To implement the run-and-tumble algorithm autonomously on the Crazyflie, we needed a Crazyflie quadcopter and these additional sensors:

The Optic Flow deck was a key sensor in achieving autonomous flight. This sensor package determines the Crazyflie’s height above the surface and tracks its horizontal motion from the starting position along the x-direction and y-direction coordinates. With the Optic Flow installed, the Crazyflie is capable of autonomously maintaining a constant height above the surface. It can also move forward, back, left, and right a set distance or at a set speed. Several other pre-programmed movement behaviors can also be chosen. This Bitcraze blog post has more information on how the Flow deck works and this post by Chuan-en Lin on Nanonets.com provides more in-depth information if you would like to read more.

The Bitcraze Multi-ranger deck provided the sensor data for obstacle avoidance. The Multi-ranger detects the distance from the Crazyflie to the nearest object in five directions: forward, backward, right, left, and above. Our threshold to trigger the “Avoid Obstacle” behavior is detecting an obstacle within 0.5 meters of the Crazyflie quadcopter.

The Prototype deck was a quick, simple way to connect the BH-1750 light intensity sensor to the pins of the Crazyflie to physically integrate the sensor with the quadcopter hardware. This diagram shows how the header positions connect to the rows of pads in the center of the deck. We soldered a header into the center of the deck, then soldered connections between the pads to form continuous connections from our header pin to the correct Crazyflie header pin on the left or right edges of the Prototype deck. The Bitcraze Wiki provides a pin map for the Crazyflie quadcopter and information about the power supply pins. A nice overview of the BH-1750 sensor is found on Components101.com, this shows the pin map and the 4.7 kOhm pull-up resistor that needs to be placed on the I2C line.

It was easy to connect the decks to the Crazyflie because Bitcraze clearly marks “Front”, “Up” and “Down” to help you orient each deck relative to the Crazyflie. See the Bitcraze documentation on expansion decks for more details. Once the decks are properly attached, the Crazyflie can automatically detect that the Flow and Multi-Ranger decks are installed, and all of the built-in functions related to these decks are immediately available for use without reflashing the Crazyflie with updated firmware. (We appreciated this awesome feature!)

## Crazyflie Firmware and ROS Control Software

Bitcraze provides a downloadable virtual machine (VM) to help users quickly start developing their own code for the Crazyflie. Our team used a VM that was modified by UW graduate students Melanie Anderson and Joseph Sullivan to make it easier to write ROS control code in the Python coding language to control one or more Crazyflie quadcopters. This was helpful to our team because we were all familiar with Python from previous work. The standard Bitcraze VM is available on Bitcraze’s Github page. The Modified VM constructed by Joseph and Melanie is available through Melanie’s Github page. Available on Joseph’s Github page is the “rospy_crazyflie” code that can be combined with existing installs of ROS and Bitcraze’s Python API if users do not want to use the VM options.

• “crazyflie-firmware” – a set of files written in C that can be uploaded to the Crazyflie quadcopter to overwrite the default firmware
• In the Bitcraze VM, this folder is located at “/home/bitcraze/projects/crazyflie-firmware”
• In the Modified VM, this folder is located at “Home/crazyflie-firmware”
• “crazyflie-lib-python” (in the Bitcraze VM) or “rospy_crazyflie” (in the Modified VM) – a set of ROS files that allows high-level control of the quadcopter’s actions
• In the Bitcraze VM, “crazyflie-lib-python” is located at “/home/bitcraze/projects/crazyflie-lib-python”
• In the Modified VM, navigate to “Home/catkin_ws/src” which contains two main sets of files:
• “Home/catkin_ws/src/crazyflie-lib-python” – a copy of the Bitcraze “crazyflie-lib-python”
• “Home/catkin_ws/src/rospy_crazyflie” – the modified version of “crazyflie-lib-python” that includes additional ROS and Python functionality, and example scripts created by Joseph and Melanie

In the Modified VM, we edited the “crazyflie-firmware” files to include code for our light intensity sensor, and we edited “rospy-crazyflie” to add functions to the ROS software that runs on the Crazyflie. Having the VM environment saved our team a huge amount of time and frustration – we did not have to download a basic virtual machine, then update software versions, find libraries, and track down fixes for incompatible software. We could just start writing new code for the Crazyflie.

The Modified VM for the Crazyflie takes advantage of the Robot Operating System (ROS) architecture. The example script provided within the Modified VM helped us quickly become familiar with basic ROS concepts like nodes, topics, message types, publishing, and subscribing. We were able to understand and write our own nodes that published information to different topics and write nodes that subscribed to the topics to receive and use the information to control the Crazyflie.

For more information, see the Bitcraze Development overview.

### Updating the crazyflie-firmware

A major challenge of our project was writing a new driver that could be added to the Crazyflie firmware to tell the Crazyflie system that we had connected an additional sensor to the Crazyflie’s I2C bus. Our team referenced open-source Arduino drivers to understand how the BH-1750 connects to an Arduino I2C bus. We also looked at the open-source drivers written by Bitcraze for the Multi-ranger deck to see how it connects to the Crazyflie I2C bus. By looking at all of these open-source examples and studying how to use I2C communication protocols, our team member Nishant Elkunchwar was able to write a driver that allowed the Crazyflie to recognize the BH-1750 signal and convert it to a sensor value to be used within the Crazyflie’s ROS-based operating system. That driver is available on Nishant’s Github. The driver needed to be placed into the appropriate folder: “…\crazyflie-firmware\src\deck\drivers\src”.

The second change to the crazyflie-firmware is to add a “config.mk” file in the folder “…\crazyflie-firmware\tools\make”. Information about the “config.mk” file is available in the Bitcraze documentation on configuring the build.

The final change to the crazyflie-firmware is to update the make file “MakeFile” in the location “…\crazyflie-firmware”. The “MakeFile” changes include adding one line to the section “# Deck API” and two lines to the section “# Decks”. Information about compiling the MakeFile is available in the Bitcraze documentation about flashing the quadcopter.

### Making additions to the ROS control architecture

The ROS control architecture includes messages. We needed to define 3 new types of messages for our new ROS control files. In the folder “…\catkin_ws\src\rospy_crazyflie\msg\msg” we added one file for each new message type. We also updated “CMakeLists.txt” to add the name of our message files in the section “add_message_files( )”.

The second part of our ROS control was a set of scripts written in Python. These included our run-and-tumble algorithm control code, publisher scripts, and a plotter script. These are all available in the project’s Github.

## Characterizing the Light Sensor

At this point, the light intensity sensor was successfully integrated into the Crazyflie quadcopter. The new code was written and the Crazyflie quadcopter was reflashed with new firmware. We had completed our initial trouble shooting and the next step was to characterize the light intensity in our experimental setup.

This characterization was done by flying the Crazyflie at a fixed distance above the floor in tightly spaced rows along the x and y horizontal directions. The resulting plot (below) shows that the light intensity increases exponentially as the Crazyflie moves towards the light source.

The light characterization allowed us to determine an intensity threshold that will only happen near the light source. If this threshold is met, the algorithm’s “Stop” action is triggered, and the Crazyflie lands.

## Testing the Run-and-Tumble Algorithm

With the light intensity characterization complete, we were able to test and revise our run-and-tumble algorithm. At each loop of the algorithm, one of the four actions is chosen: “Run”, “Tumble”, “Avoid Obstacle”, or “Stop”. The plot below shows a typical path with the action that was taken at each loop iteration.

## Flight Tests of the Run-and-Tumble Algorithm

In final testing, we performed 4 trial runs with 100% success locating the light source. Our test area was approximately 100 square feet, included 1 light source, and 2 obstacles. The average search time was 1:41 seconds.

## Lessons Learned

This was one of the best courses I’ve taken at the University of Washington. It was one of the first classes where a robot could be incorporated, and playing with the Crazyflie was pure fun. Another positive aspect was that the course had the feel of a boot camp for learning how to build, control, test, and improve autonomous robots. This was only possible because Bitcraze’s small, indoor quadcopter with optic flow capability made it possible to safely operate several quadcopters simultaneously in our small classroom as we learned.

This development project was really interesting (aka difficult…) and we went down a few rabbit holes as we tried to level up our knowledge and skills. Our prior experience with Python helped us read the custom example scripts provided in our course for the ROS control program, but we had quite a bit to learn about the ROS architecture before we could write our own control scripts.

Nishant made an extensive study of I2C protocols as he wrote the new driver for the BH-1750 sensor. One of the biggest lessons I learned in this project was that writing drivers to integrate a sensor to a microcontroller is hard. By contrast, using the Bitcraze decks was so easy it almost felt like cheating. (In the nicest way!)

On the hardware side, the one big problem we encountered during development was accidentally breaking the 0.5 mm headers on the Crazyflie quadcopter and the decks. The male headers were not long enough to extend from the Flow deck all the way up through the Prototype deck at the top, so we tried to solder extensions onto the pins. Unfortunately, I did not check the Bitcraze pin width and I just soldered on the pins we all had in our tool kits: the 0.1 inch (2.54 mm) wide pins that we use with our Arduinos and BeagleBones. These too-large-pins damaged the female headers on the decks, and we lost connectivity on those pins. Fortunately, we were able to repair our decks by soldering on replacement female headers from the Bitcraze store. I wish now that the long pin headers were available back then.

In summary, this course was an inspiring experience and helped our team learn a lot in a very short time. After ten weeks working with the Crazyflie, I can strongly recommend the Crazyflie for robotics classes and boot camps.

## Links to Project Files

Team’s Research Poster: https://github.com/thecountoftuscany/crazyflie-run-and-tumble/blob/master/documents/Project-Final_Poster.pdf

Github Link courtesy of Nishant Elkunchwar: Crazyflie-Run-and-Tumble

YouTube Links courtesy of Nishant Elkunchwar: Crazyflie locates Light, Simulation of Run and Tumble Algorithm in PyGame

It has been a while since we have updated you all on the AI deck. The last full blogpost was in October, with some small updates here and there. It is not that we have not focused on it at all; on the contrary… this has been a high priority project for a while now. It is just quite a complex board with a lot of bells and whistles, which can be challenging to work with sometimes so early in development, something that our previous intern can definitely agree on. So therefore we rather wanted to wait until we were able to make sufficient progress before we gave you an update… and so we have!

Together with Greenwaves technologies we have been trying to get the SDK of the GAP8 chip on the AI deck stable enough for an early release. The latest release of the SDK (version 3.4) has proved itself to work with relative ease on the AI deck after extensive testing. Currently it is possible to use OpenOCD for flashing and debugging, and it supports most commonly available debuggers with a jtag connector. In the upcoming weeks both of Bitcraze and Greenwaves will test and try out all examples of the SDK on the AI deck to make sure that everything is still compatible. Also the documentation will be extended as well. As there is so much to document, it might be difficult to catch all of it. However, if you notify us and Greenwaves on anything that is missing once the AIdeck is out, that will help us out to catch the knowledge gaps.

The AI deck also contains the ESP-based NINA module for establishing a WiFi connection. This enables the users to stream the video stream of the AI deck onto their computers, which will be quite an essential tool if they would like to generate their own image database for training the CNNs for the GAP8 (and it happens to also be quite practical for debugging by the way!). Currently it is required to set credentials of your local WiFi network and reflash the AI-deck to be able to connect and streaming the images, but we are working on turning the Nina into an access-point instead so no reflashing would be required. We hope that we will be able to implement this before the release.

We are also trying out to adjust applications to make suitable of the AI deck. For instance, we have adapted Greenwaves’ face-detector example to use the image streamer instead of the display available on the GAPuino boards. You can see a video of the result here underneath. Beware that this face-detector is not based on a CNN but on HOG descriptors, so it only works in good conditions where the face is well lit. However, it is possible to train a CNN to detect faces in Tensorflow and flash this on the AI deck with the GAPflow framework as developed by Greenwaves. At Bitcraze we haven’t managed to try that out ourselves ( we are close to that though!) but at least this example is a nice demonstration of the AI deck’s abilities together with the WiFi-streamer. This example and more testing code can be found in our experimental repo here. For examples of GAPflow, please check out the examples/NNtool section of the GAP8 SDK.

For some reason WordPress has difficulty embedding the video that was supposed to be here, so please check https://youtu.be/0sHh2V6Cq-Q

Seeing how the development has been progressing, we will be comfortable to say that the AI deck could be ready for early release somewhere in the next month, so please keep an eye out on our website! We will continue to test the GAP SDK’s stability and we are very thankful for Greenwaves Technologies with their help so far. We will also work on getting-started guides in order to get acquainted with the AI deck, supplementing the already existing documentation about the GAP8 chip.

Even-though the AI deck will soon be ready for early release, this piece of hardware is not for the faint-hearted and embedded programming experience is a must. But keep in mind that the possibilities with the AI deck are huge, as it will be mean that super-edge-computing on a 30 gram flying platform will be available for anyone. It will all be worth it when you have your Crazyflie flying autonomously while being able to recognize its surroundings :)

We have a guest blog post this week from Christopher Banks at Georgia Tech, where he tells us about their work with the Robotarium. Enjoy!

## Multi-Agent Aerial Robotics

In the GRITS Lab  we focus on autonomous control and coordination of multi-robot systems with applications in – but not limited to – optimal control, constraint-based control, and hardware development. We are home to the Robotarium [1], a remotely accessible swarm robotics testbed that is free for anyone around the world to use for academic and educational purposes. We have integrated Crazyflies into the Robotarium as the main vehicle for aerial robot swarms due to their small size, quiet operation, and high maneuverability . Also, due to their low inertia, they pose minimal harm to their surroundings if system failures occur. Their small size and robust nature are well suited for flying in an indoor testbed like the Robotarium. As we work towards extending the operation of the Crazyflies in the Robotarium to external users, we encountered some important research questions: How do we guarantee the quadcopters remain “safe” (undamaged) while minimizing modifications to user inputs? How do we develop an easy to use interface for external users, with experience ranging from novice to expert? What commands can be used by external users to control a swarm of robots?  This post will briefly describe the ongoing research aimed at solving these questions.

### Safety Guarantees

To ensure hardware safety while flying experimental algorithms we have developed Control Barrier Functions (CBFs) for quadcopters, allowing users to give nominal control inputs while obeying some safety constraints for the system (e.g. collision-free trajectory following). In the video below, we give four Crazyflies the commands to fly in a circle. A fifth Crazyflie is then told to fly to waypoints that will intersect the circle and attempt to collide with the circling quadcopters. Using CBFs a central controller can modify the inputs given to Crazyflies near collision to ensure safe velocity commands that are close as possible to the user intended control [2] . These CBFs can also be designed to ensure safety by bounding the quadcopters to a designated region of the testbed, giving additional safety constraints by protecting areas outside of the motion capture system during flights.

### User Interfaces

We have also used the Crazyflies to understand how remote users can best interact with the Robotarium both at the interface level and in planning. One project involved studying the effectiveness of graphical user interfaces (GUIs) on swarm robotic control. Two GUIs were developed with different interaction modalities. The GUIs were designed to map user inputs to a set of hoops placed in the Robotarium. One GUI (shown in Fig. 1) provided users the ability to draw paths through a touchscreen interface on a two-dimensional map and then map those inputs to trajectories for a team of robots. The other GUI (illustrated in Fig. 2) allowed users to input a sequence of desired hoops for a team of robots and execute trajectories based on the input.

### Multi-Agent Planning

In planning, we looked at how multi-agent planning can be approached using high-level specifications. These high-level specifications allow users to develop plans requiring groups of robots to visit regions of interest (see Fig. 3) and trajectories are generated automatically. To represent these specifications, we use a logic formalism known as temporal logic to encode a preferred sequence of plan execution. As an additional step, users could include constraints on the trajectory by minimizing a cost using stochastic sampling. For more details, see the attached video demonstrating task allocation in a fire-fighting scenario.

## Future Directions

As we continue to expand the capabilities of the Robotarium we are looking into how to develop long term autonomy for the Crazyflies. This includes autonomous charging as well as remote access for the lab and other users. We hope to use the Lighthouse system as a method for long term tracking since the Crazyflie will know its position instead of relying on passive tracking from a Vicon system. Our plans also include a lab-based simulator for in house projects related to the Crazyflies as well as updating our system to incorporate Crazyswarm to make control of the Crazyflies easier in implementation. In addition to this, in order to accommodate unknown users, we will have to figure out a control scheme that encourages use from a wide variety of users ranging from novices in quadcopter control to experts. We’ll keep Bitcraze updated on the Robotarium’s progression towards fully autonomous aerial swarms!

1. Robotarium Article: https://ieeexplore.ieee.org/document/8960572
2. CBFs for Quadcopters: https://ieeexplore.ieee.org/document/7989375

Accurate indoor localization is a crucial enabling technology for many robotic applications, from warehouse management to monitoring tasks. Ultra-wideband (UWB) localization technology, in particular, has been shown to provide robust, high-resolution, and obstacle-penetrating ranging measurements. Nonetheless, UWB measurements are still corrupted by non-line-of-sight (NLOS) communication and spatially-varying biases due to doughnut-shaped antenna radiation pattern. In our recent work, we present a lightweight, two-step measurement correction method to improve the performance of both TWR and TDoA-based UWB localization.  We integrate our method into the Extended Kalman Filter (EKF) onboard a Crazyflie and demonstrate a closed-loop position estimation performance with ~20cm root-mean-square (RMS) error.

## Methodology

UWB measurement errors can be separated into two groups: (1) systematic bias caused by limitations in the UWB antenna pattern and (2) spurious measurements due to NLOS and multi-path propagation. We propose a two-step UWB bias correction approach exploiting machine learning (to address(1)) and statistical testing (to address (2)). The data-driven nature of our approach makes it agnostic to the origin of the measurement errors it corrects.

### (1) Neural Network Bias Correction

The doughnut-shaped antenna radiation pattern causes the relative poses of anchors and tags to have a noticeable impact on the received signal power, which leads to systematic, predictable biases.  To empirically demonstrate the systematic measurement errors resulting from varying the relative pose between anchors and tags, we placed two DWM1000 UWB anchors at a distance of 4m and collected both TWR and TDoA UWB range measurements for the UWB tag mounted on top of a Crazyflie spinning around its own z-axis.

We choose to leverage the nonlinear representation power of neural networks to learn the systematic bias which only depends on anchor-tag relative poses. Considering the limited onboard computation power, we select a fully connected neural network with 50 neurons in each of two layers with ReLU activation. To represent the relative pose between the UWB tag and anchors, we select the relative distance ∆p and roll, pitch, and yaw angles of the quadcopter as the input features x for the network. As we used fixed anchors, we do not include their poses as inputs (this level of generalization is left for future work). Given sufficient training data, the spatially-varying measurement bias can be described by a nonlinear function b=f(x) captured by the trained neural network.

### (2) Outlier (Spurious Measurements) Rejection

Besides our learning-based bias correction, we use a quadcopter’s dynamic model to filter inconsistent UWB range measurements. Given the estimated velocity v and maximum acceleration amax, we can compute the maximum distance dmax a quadcopter can cover during time ∆t. Based on this information, we can reject unattainable measurements before fusing them into the EKF by comparing the measurement innovation with dmax

Moreover, we use a statistical hypothesis test to further classify potential outlier measurements. Since the measurement innovation vector is assumed to be distributed according to a multivariate Gaussian distribution, the normalized sum of squares of its values should follow a Chi-square distribution. We use the Chi-square hypothesis test to determine whether a measurement innovation is likely coming from this distribution.

## Data Collection and Training

We use a Crazyflie 2.0 quadcopter and the Loco Positioning System (LPS)’s UWB DW1000 modules as our research platforms. Our calibration approach runs on the Crazyflie STM32 microcontroller within the FreeRTOS real-time operating system. We equipped a cuboid flying arena with 8 UWB anchors, one for each vertex. The anchor positions were measured using a Leica total station theodolite.

For all experiments, the ground truth position of the Crazyflie was provided by 10 Vicon cameras. The neural network was trained using PyTorch. To perform inference on the Crazyflie’s microcontroller, we re-use PyTorch’s trained weights in a plain C re-implementation. Since the DW1000 modules in the LPS provide UWB measurements every 5ms, the neural network inference runs at 200Hz during flight as well. Our outlier rejection method is also implemented in plain C and merged with the onboard EKF.

## Close-loop Position Estimation Performance

We demonstrate the position estimation and close-loop performance of the proposed methods by flying a Crazyflie quadcopter along planar and non-planar circular trajectories (which were not among the trajectories used for training). A comparison between the estimation error of (A) the UWB localization estimate enhanced with outlier rejections and (B) the estimated enhanced with both outlier rejection and neural network bias compensation is conducted in our experiments for both TWR and TDoA2 modes. We repeated all of our experiments 10 times with a target velocity of 0.375m/s. The quadcopter trajectories during these flight tests are displayed in the following plots.

The distributions of the RMS estimation errors are summarized into a box plot. TWR-based ranging results in better localization performance than TDoA. However, we observe that, with our neural network bias compensation, the average RMS error of TDoA localization is around 0.21m, which is comparable to that of TWR-based localization (~0.19m). Thanks to the neural network bias compensation, the average reduction in the RMS error is ~18.5% and 48% for TWR and TDoA, respectively. Most notably, this result suggests that bias compensation might help closing the performance gap between TWR- and TDoA-based localization.

## Outlook

In this work, we presented a two-step methodology to improve UWB localization—for both TWR- and TDoA-based measurements. We used a lightweight neural network to model and compensate for pose-dependent and spatially-varying biases and an outlier rejection mechanism to filter spurious measurements. Through several real world flight experiments tracking different trajectories, we showed that we are able to improve localization accuracy for both TWR and TDoA, granting safer indoor flight. In our future work, we will include the anchors’ pose information to allow our method to further generalize to previously unobserved indoor environments, with different anchor configurations.

The authors are with the Dynamic Systems Lab, Institute for Aerospace Studies, University of Toronto, Canada, and affiliated with the Vector Institute for Artificial Intelligence in Toronto.

Feel free to contact us if you have any questions or ideas: wenda.zhao@robotics.utias.utoronto.ca. Please cite this as:

<code>@article{wenda2020learning,
title={Learning-based Bias Correction for Ultra-wideband Localization of Resource-constrained Mobile Robots},
author={Wenda Zhao and Abhishek Goudar and Jacopo Panerati and Angela P. Schoellig},
journal={arXiv preprint arXiv:2003.09371},
year={2020}
}</code>

The Lighthouse V2 implementation has been simmering away for a long time in the Bitcraze kitchen and in this blog post we will give you an update on the current status and what is remaining for a full release of this tasty dish.

We believe we have solved most of the major technical hurdles (last famous words) on the way to a working implementation that uses Lighthouse V2 base stations for positioning, now it is mostly work to implement the functionality that is remaining. As described in this post we now have a new FPGA binary that has the ability to decode both V1 and V2 base stations, and this was a major step forward. This new binary is used in the Crazyflie firmware master branch, and if the Lighthouse deck is used with the latest Crazyflie firmware, the new FPGA binary will automatically be flashed to the deck.

## What has changed?

The new FPGA binary uses a different UART protocol to communicate with the Crazyflie. This protocol has been implemented in the firmware and hopefully there is no functional difference compared to the previous FPGA binary when using Lighthouse V1 base stations.

We have added a first version of Lighthouse V2 base station decoding, but it is still a bit limited. As a start we decided to “emulate” V1 base stations to be able to reuse as much of the existing code as possible. For now we support only 2 base stations and they must use channel 1 and 2 (used to be called modes). The V2 angles are transformed into V1 angles and fed into the old positioning logic and are handled exactly the same way as before. Even though this works, it is not the optimal solution and we hope to be able to refine it later on.

We have also written a python script to estimate base station geometry (positions and orientation) using the Lighthouse deck. This removed the requirement to use software from Steam which should simplify the set up process. Please see the (still limited) documentation. Note that this calibration method only supports the basestation V1… for now!

There is a lot of code that has been modified and the FPGA implementation is completely new, it is not unlikely that there is functionality that is unstable or broken, or configurations that are not supported. If you happen to notice any bugs, please let us know!

## What is remaining?

The functional areas that needs to be implemented or cleaned up before we leave the Early Access stage is the following:

### Calibration data

The calibration data is embedded in the modulated light from the base stations and describes imperfections from the manufacturing process for each individual. This data is not read yet for V2 and will increase the precision when available.

### Support for more than 2 base stations

Lighthouse V2 base stations are designed for systems with more than 2 base stations. The Crazyflie firmware needs to be extended for this functionality to work, including handling of geometry data, logging, memory management and some other bits and pieces.

### Native V2 positioning

The angles from the V2 base station should be fed directly into the kalman filter for positioning, instead of first being transformed into V1 angles. This will increase robustness and reduce data loss.

### Client support

We want to add a tab in the python client where a Lighthouse system can be monitored, configured and managed. It should, for instance enable the user to configure and visualize base station geometry.

### FPGA binaray management

Currently the FPGA binary is included in the Crazyflie firmware and it is automatically uploaded to the deck when booted. This is not a viable long term solution and we hope to be able to find a more generic way of handling deck binaries.

## Conclusions

As can be seen, there is still quite some work to be done before the Lighthouse V2 stew is ready to be served, but we are definitely starting to smell some nice flavours from the kitchen!

Finally a view from Kristoffer’s home lab, currently in the summer house. Three base stations are set up as a Fun Friday hack to see what it would take to use more than 2. Luckily it did not take too much time to get this to work :-)

Many people in the world have now settled in the reality of working from home. We have also taken precautions ourselves by not go to our office as normal and only ship out packages a few times per week instead of every day (see this blogpost). This also means that we do not have full access to our lab with all our equipment and positioning systems in our big 10 x 10 meter flight lab at the office. In this blogpost we will show how we manage to keep on developing and flying, even in the current situation.

### In(light)house positioning

Currently we started to use the Lighthouse positioning system to setup up the remote home lab at our houses. As of recent additions to the Crazyflie firmware, it has been made easy to get the geometry data from the base station. Now the only items we need for indoor flight are just two (or only one) lighthouse basestations V1’s and a Crazyflie, and that is it! There is no need for an HTC Vive headset or hub, or third-party software like SteamVR and the setup is finished in 2 minutes! Check out the new documentation here if you want to know more about the new setup of the lighthouse positioning system.

Also, we recently got a very primarily version of the lighthouse V2 working (see here) and we of course want to keep the momentum going! We will be working on full compatibility from our homes so stay tuned. For now, see this video of the Crazyflie flying with just a single base-station, taken from one of our team-member’s home lab.

### Remote Lecture Hall and Practicals

We were invited by Dario Floreano and Fabrizio Schiano from the EPFL-LIS laboratory to do a lecture for the ‘Aerial Robotics’ Course as part of EPFL’s Master’s program in Robotics. Due to the virus, we had to cancel our trip to go there physically… but luckily we were able to do the lecture remotely anyway!

The lecture consists of two parts. In the first hour we mostly explained about the Crazyflie ecosystem, hardware and sensors. In the second hour we focused on how the stabilization module worked, including the controllers and the state estimation. During both sessions, we alternated between the theory slides with actual hands-on demos. The lighthouse positioning system was setup in a kitchens, so that we were able to show full flights and practicals with the Crazyflie. At the end there was also the push-demo with just the flowdeck and multiranger, which didn’t use any external positioning at all.

The lectures can be found below and the documentation has been updated as well with the covered material (see here). Be sure to check out the controller tuning presented in part 2 of the lecture (25:00 – Cascaded PID controller).

## Other Home labs

We are happy to announce that we have gotten Crazyflie 2 to fly autonomously using the Lighthouse deck and Lighthouse V2 base-stations. This was a very requested features, and while this is not stable and ready to use yet, it is a great milestone toward Lighthouse V2 support.

There exists two incompatible versions of the Lighthouse positioning system. Version 1 was released with the original HTC Vive VR system. In this system base-station are using two rotating laser beam that sweeps the room, one horizontal and one vertical, and an omnidirectional synchronization flash to allow IR light receiver to be located in the room. One limitation of this version is that up to two base-station can be used and no more, this is mainly due to the fact that beam identification is done using a TDMA scheme: base stations switch-on their laser in a dedicated time-slot one after each-other and adding more time slots for more base-stations will greatly reduce the update rate of the system.

Lighthouse V2, was released with the HTC Vive PRO headset and is also used by the Valve Index. The big change is that laser sweeps now carries modulated data and that there is only one rotor with two angled slit instead of the two rotors for V1. The V2 sweep data is described as ‘Sync on beam’ and contains timing information of how long it has been since the synchronization event (ie. when the rotor crossed 0 degree). The sweep data also allows to identify the base-station that has transmitted the sweep. This removes the need for an omni-directional synchronization pulse and allows more than two base-station to operate at the same time in the same space, since their sweeps can now be identified and timed.

The lighthouse V2 system is very elegant and scalable. However, actually decoding the signal from the sweeps has taken a lot of time since it is not documented and we needed to find-out what the encoding actually was. There has been effort on the internet to understand how the system worked, the most useful one is this github ticket that goes from raw data acquisition to fully unlocking the beam encoding.

I have been working on-and-off for a long time on making an FPGA design for the lighthouse deck to acquire and decode Lighthouse V2. The main blocking point until now was that I had not been able to reliably acquire useful signal from the system in order to allow real-time decoding on the Crazyflie. Added to that, there was some inconsistency between what we though the system was doing and what we could gather from the base-stations debug console. Recently though, the last piece of the puzzle, was to discover that the beam encoding was not Manchester, as we though, but Bi-phase mark code FM1 (BMC). Once this decoding was used everything made sense and worked.

Added to that, I started using SpinalHDL instead of raw Verilog to write the FPGA design which allows for much quicker iteration, much less frustration, and it also allowed me to easily make the design multi-clock which is required to decode the BMC signal: the beam decoder runs at 48MHz, and the rest of the system works at 24MHz. This design is required since the FPGA we use in the lighthouse deck is not fast enough to run everything at 48MHz.

The result, is a new FPGA firmware for the lighthouse deck that receives, identify and decode Lighthouse V2 sweep signal and send them over to the Crazyflie. The Crazyflie still has a little pulse packing to do (putting together pulses from a single sweep received on multiple sensors) and then can use pulse timing information to calculate azimuth and elevation at which the base-station sees the Crazyflie. This information is the same as the one we get from Lighthouse V1 and so the same algorithm can be used to calculate the Crazyflie position.

I hacked a proof of concept was this last fun Friday and it flies!

If anyone is curious the code for this demo has been uploaded as an out-of-tree driver and the code for the FPGA parts is already in the lighthouse-fpga project. The current Crazyflie code is too incomplete to be usable, but it is a nice starting point if anyone wants to play with Lighthouse V2 and the Crazyflie right away ;-).

As a side note, the Bitcraze team will shrink temporary as I, Arnaud, will go in parental leave until mid-August. I look forward to this new adventure and I trust the lighthouse V2 development and the forum will be in good hands in my absence.

Happy holidays to all our users, community members and friends! We are happy to announce our 2019 Christmas video which we have made in collaboration with Ben Kuper! It is starring 7 Crazyflies, the lighthouse positioning system, our office Christmas tree and a whole lot of holiday spirit, so go ahead and take a look!

Here are some words from Ben how it was to work on this year’s Christmas video at our office:

Coming to Bitcraze’s HQ and working with them has been once more a wonderful experience, technically and humanly ! The main goal of this session was to test and implement the new lighthouse tracking system in the tool suite I’m creating, and it was an amazing surprise to witness for real the uncanny stability of the drones when they’re on lighthouse tracking !

Of course, my first reaction was to push the limit and see what can be done with this new power, this is why I created this choreography : to see what can be done in a limited amount of time (1 and a half day to create the full choreography, the official video shows the first part only), and trying to go at the limit of the current possibilities. As the team was working on occlusion recovery, we decided to have the drone fly around the tree as a fun test, and it works !

In the new year we will have a followup blog-post going into detail on how exactly we made this video. Until then, happy holidays and have an awesome new year!