Perception platform for autonomous vehicles
Project ID: AEE-2017-16
Students: Ville Kukkonen, Amin Modabberian, Pietu-Pekka Roisko, Pyry Viita-aho, Ilmari Vikström
Project manager: Ville Kukkonen
Instructor: Arto Visala
Other advisors: Andrei Sandru
Starting date: 5.1.2017
Completion date: 22.5.2017
The objective of the project was to further develop the perception capabilities of an all-terrain vehicle fitted with a LiDAR, an omnidirectional camera and optical encoders for wheels and steering wheel. The basic groundwork of communicating with most of the sensors via ROS had been done in previous projects, and our task was to do the missing integration and calibration, as well as to lay the foundations for autonomous capabilities such as path planning and obstacle avoidance.
Summary of results
The omnidirectional camera was integrated to ROS, and the spherical image it produced was transformed to panoramic form. A kinematic model was implemented, utilizing the wheel and steering encoders, and combined with results from the INS to provide fused localization information. A SLAM solution was implemented and calibrated, using the LiDAR and the fused localization.