|
2011 SparkFun Autonomous Vehicle CompetitionPromises to test ahead of time were shamelessly broken this year, as I went to the 2011 SparkFun Autonomous Vehicle Competition with an unfinished robot. The vehicle was much improved over last year, with more robust inertial-navigation capability along with laser-rangefinding obstacle detection. Unfortunately, a persistent problem with the serial communication between the image processing computer and the control hardware prevented the system from working. Although I did not end up running, it's hard to leave an event that featured a recklessly-autonomous 7-foot dinosaur robot without having a great time. The competition this year was fantastic, and much of my report is dedicated to comparing the various approaches taken with the ground vehicle designs. I also include a description of my robot - with particular emphasis on the hardware and control system design. |
|
|
2010 SparkFun Autonomous Vehicle CompetitionAutoDrift debuts at the 2010 SparkFun AVC in Boulder, CO. The competition requires robots autonomously navigate the SparkFun building, either by air or on the ground. The competition summary page details my vehicle, control design, and competition experience. Not much in the way of racing success, but the AutoDrift platform is maturing and should be ready for autonomous drift trials in summer 2010. AutoDrift currently performs cross-track waypoint navigation between up to 10 waypoints, and stabilizes the inner-loop dynamics using yaw rate tracking or lateral acceleration tracking. Longitudinal speed tracking controller tracks a desired GPS speed, with high- speed segments between waypoints and low-speed segments on approach to waypoints and during large turn commands. A crude obstacle avoidance controller steers away from identified obstacles and reduces speed when obstacles are detected. The obstacle avoidance is limited to relatively low speeds due to the limited range of the ultrasonic proximity sensors. |
Office M&M Dispenser MazeShort video of my M&M dispenser project (full version), which was made to compete with other candy dispensers around our office (M&M, mint, and jellybeans). My goal was to make a dispenser that was somewhat more entertaining than reaching into a jar, but also reduces the large M&M flux caused by greedy handfuls. The dispenser maze requires the prospect candy-eater to tilt the maze table and guide the M&M through the sequence to the Win Bucket. There are a number of open wall segments on the maze, beyond which lie the fail ramps that roll the M&M down to the fail bucket. While the M&Ms are still accessible in the fail bucket, candy-eaters must face public scorn and shunning from colleagues if they eat from there. Control of the maze is provided by a hand-held controller, which uses 8-bit Nintendo POWER/RESET buttons for M&M dropping, and an analog Playstation joystick for table tilting. An ATTiny 861 microcontroller reads the analog position of the joystick and generates PWM commands to drive the pitch and roll tilt servos, which are connected to the table and provide roughly +/- 10 deg tilt angle. |
Honey Bee Helicopter versus Potato CannonVideo documents the environmentally friendly disposal of an old fixed-pitch E-Sky Honey Bee R/C helicopter. An organic compound is used to facilitate the mechanical breakdown of the helicopter components and spread them over a wide area of an environmental impact research facility. Clearly, the solution to pollution is dilution. The authors of the study would like to acknowledge the contributions of Chuck from Von's supermarket, who helped identify the appropriate tubers that have sufficient mass, cohesion, and workability to function as the dispersion catalyst. |
Slow Motion Mini-Z CrashesOverhead view of carnage during our weekly office Mini Z races. We had an open-intersection in the last few turns of our track, with a partial figure-8 right before the finish line. None of the crashes were intentional (right guys?), but we were rolling footage, just in case. Filmed using a Photron SA-3 high-speed camera at 1000 frames per second. The video is a cropped version of the full-frame video, which shows more of the moments prior to and following the impacts. |
Honey-Bee Helicopter Obstacle Course RacingManeuver Sequence: 1 - Take-Off from Pad 2 - Blow origami boulders off table 3 - Touch-and-go on three pedestals 4 - Table loops with touch-and-goes 5 - Table leg figure-8s 6 - Balloon popping 7 - Ladder fly-through 8 - Landing on Pad |
Aerobatics Flight Data
Aerobatic maneuvers of an instrumented R/C aircraft demonstrate stick inputs
and dynamic response during agile flight. The R/C Flash 3D EPP aircraft is
equipped with instrumentation to measure accelerations, angular rates,
pitot-static pressure, GPS, and servo motion at 100Hz. The data is recorded
to on-board flash memory and is animated and synchronized to the external
video in post-processing. DataNinja is used to measure and record the analog
signals generated by the IMU, NinjaSense. |
Mitigating the Effect of Atmospheric Turbulence: Toward More Useful MAVsRMIT University, Shifted Dynamics, and Monash
University conducted a series of UAV flight tests in a large, wind engineering tunnel.
The video presents highlights of tests, which incorporate parametric variation of the
aircraft configuration and geometry to assess the suitability for flight in relatively
large levels of turbulence. These tests are part of ongoing research into strategies for
mitigating the effect of turbulence on the flight of various UAV platforms. |
![]() |
Recent developments in drift research
were presented at the 2006
SAE World Congress in Detroit, MI, April 3-6, 2006. The
paper summarizes the variations in tire force and moment equilibrium
at various drifting angles.
SAE Paper 2006-01-1019 (click to download PDF manuscript) Abstract
|
![]()
Drift testing with 4 in-line accelerometers, sideslip camera, and
steering angle sensor. March 12, 2006 |