IGVC

FIT IGVC (igvc.org for more info)
Intelligent Ground
Vehicle Competition
Goals and
Competition
• Navigate a random course
• Map a course in real time
• Detect lanes and obstacles reliably
• Localize the robot accurately
• Create an interoperable, adaptable
software architecture
Where are we now?
• Functioning line detection
• Software baseline and
communication framework
• Novel motion planner
• Readings from necessary sensors
• Working on integration
• Code is in Java, C/C++
• We use IntelliJ and CLion for
development
• RabbitMQ & JSON used across
the platform
• Modular software architecture
that allows plug and play
behavior
Line Detection & Obstacle Detection
• Stereoscopic camera based line detection
• Rapid line detection using entirely GPU
based software
• High accuracy line detection
• In process on obstacle detection
Motion
Planning
• Probabilistic motion planning
• Dynamic re-planning
• Real time course mapping
Sensors & Platforms
In Use
Currently:
• ZED Stereoscopic Camera
• Jetson TX1
• VectorNav 200 INS
Future:
• Industrial grade LiDAR
The equipment we use is on par with industry tools
or actually in use by industry
What is there to do?
• Improve the motion planner
• Implement true localization
• Simulations
• Implement interoperability
• Explore the full capabilities of the ZED camera
• LiDAR based obstacle detection
• Software testing
• Improve the communication framework
• General software engineering concepts
• Producing an awesome robot
Support for the Project
• Three professors
• Dr. Jensen (Mechanical Engineering)
• Dr. Smith (Computer Engineering)
• Dr. Gibbs (Systems Engineering)
• Graduate students working on a scaled up version of the project
• FabLab
• Working on grants to provide additional funding to the project
• Industry support from Magna & Polaris
The Team
What we are looking for
Contact to get involved
• 3 -5 Mechanical Engineers
Dr. Jensen ([email protected])
Will Nyffenegger
([email protected])
• 5 Computer / Electrical Engineers
• 3 – 4 Computer Scientists