Friday, January 16, 2009
Three-dimensional Sensor Modeling and Stimulation for the DARPA Urban Challenge
MARC U*STAR Fellow
Three-dimensional Sensor Modeling and Simulation for the DARPA Urban Challenge Simulation is the process of generating the behavior of the model using computing systems. The autonomous vehicle named ÂAliceÂ, which students and faculty at Caltech worked on for competition in the DARPA Urban Challenge, is equipped with various sensors, including Laser Detection and Ranging [LADAR] sensors and Radio Detection and Ranging [RADAR] sensors. Using the robotics simulation platform Gazebo, LADAR sensor models with artificial noise were implemented into a three-dimensional closed-loop simulation with simulated traffic. Simulation environments were modeled after actual field test environments used with Alice. Through statistical comparisons between simulated data and real data taken from field tests of Alice, the sensor models yielded data similar to those of the actual sensors during field tests. With simulations that generate statistically accurate sensor data, teams working on sensing algorithms for Alice will be able to adequately examine the performance of their algorithms in simulation as well as during field tests.