The objectives of the project were to navigate the vehicle around a simulated neighborhood to hit twelve goal points placed throughout the simulation with a time constraint of ten minutes.
Sponsor
Ryan Moyer
Team Members
Zikun Zheng | Courtland Corrente | Alex Huang | Daniel McVaney | Hoang Pham | Kyle Delhagen | | | | | |
Project Poster
Click on any image to enlarge.
Project Video
Project Summary PDF
Project Summary
Overview
The Shell Eco-Marathon Autonomous Programming Competition is a virtual international competition with a goal of designing and programming an autonomous simulated vehicle in an unreal engine environment. The Robot Operating System was used to control the vehicle using Python code. During the design process, the team had to consider path planning algorithms, as well as velocity and steering controllers. The team also faced a time constraint, as there was only three weeks left in the competition when we entered.
Objectives
The objectives of the competition were to navigate the vehicle around a simulated neighborhood to hit twelve goal points placed throughout the simulation with a time constraint of ten minutes. Our score would be determined by number of goal points hit, the energy the car consumed, the percentage of CPU used to run the code, the distance the car traveled, and the time it took the car to travel this distance.
Approach
-Determined scoring criteria of competition outlined on Shell Eco-marathon website.
-Shell-provided code was analysed to learn how to control the simulation.
-Contacted outside help to aid in the coding effort due to time constraints.
-Path planning algorithm developed using Dijkstra’s algorithm in MATLAB.
-Simulation was generated and limits on cars speed and handling were tested.
-Velocity and steering controllers coded in Python and tested in simulation.
-Feedback from shell after code submission showed that distance needed to be reduced.
-U-turns were implemented in path planning and reduced distance by 300 meters.
-Team placed 13th in the world and 2nd in North and South America.
-After competition path planning was converted to python code using an A* algorithm.
-Burn and Coast velocity control implemented as an attempt to reduce energy consumption.
-LiDAR capabilities utilized to implement object detection and avoidance.
-Full run of path completed with all 12 goal points hit.
Outcomes
-Developed an autonomous vehicle that can successfully compete in the shell Eco-Marathon Autonomous Programming Competition.
-Ranked 13th worldwide and 2nd in North and South America among the teams that participated.
-Documented code and procedures to assist future teams in competing in the competition more effectively.