MIT RACECAR Course using ROS

| No Comments | No TrackBacks
From Michael Boulet via ros-users@

We would like to announce the recent completion of the Rapid Autonomous Complex-Environment Competing Ackermann-steering Robots (RACECAR) class. RACECAR is a new MIT Independent Activities Period (IAP) course focused on demonstrating high-speed vehicle autonomy in MIT's basement hallways (tunnels). The MIT News Office published an overview article with video at: https://newsoffice.mit.edu/2015/students-autonomous-robots-race-mit-tunnels-0406 . The course website is: http://racecar.mit.edu/ .

 

Instructors provided student teams with an model car outfitted with sensors, embedded processing, and a ROS-based software infrastructure. The base platform is a Traxxas 1:10-scale radio-controlled (RC) brushless motor rally car that is capable of reaching 40+ mph speeds. The sensor suite consists of a Hokuyo 10m scanning lidar, Pixhawk's PX4Flow optical flow camera, a Point Grey imaging camera, and SparkFun's Razor inertial measurement unit. Control and autonomy algorithms are processed on-board with an embedded NVIDA Jetson TK1 development kit running the Ubuntu Linux operating system with "The Grinch" custom kernel. The TK1's pulse width modulation (PWM) output signals drive the motor electronic speed controller and steering servomotor, bypassing the RC receiver.

 

The system uses the Robot Operating System (ROS) framework to facilitate rapid development. Existing ROS drivers (urg_node, razor_imu_9dof, pointgrey_camera_driver, and px4flow_node) receive data from the sensors. The model car's throttle and steering signals are commanded with a new ROS driver interface to the kernel's sysfs-based PWM subsystem. Students develop software and visualize data through a wireless network connection to a virtual machine running on their personal laptops.

 

Given the hardware platform and basic teleoperation software stack, teams of 4 -5 students prototype autonomy algorithms over an intense two week period. Students are invited to explore a variety of navigation approaches, from reactive to map-based control. At the end of the class, the teams' solutions are tested in a timed race around a closed-circuit course in MIT's tunnels. In January 2015, three of four teams reached the finish line with the winning team's average speed exceeding 7mph.

 

We would like to thank the many contributors to ROS and, in particular, Austin Hendrix for hosting the armhf binaries at the time.

 

RACECAR Instructors

No TrackBacks

TrackBack URL: https://www.ros.org/mt-tb.cgi/1736

Leave a comment

Find this blog and more at planet.ros.org.


Monthly Archives

About this Entry

This page contains a single entry by Tully Foote published on May 6, 2015 9:28 AM.

ROS Summer School in Aachen (August) was the previous entry in this blog.

New driver for DepthSense DS325 3D Camera is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.