Robot MacGyver Project
Research project developing a generalized framework for robot tool improvisation using affordances. Complex reasoning allows robots to “think outside the box” using available items to improvise in unforeseen environments.
- Advised by Prof. Henrik Christensen.
- Developed recognition and use of object/tool affordances.
- Explored the use of behavior trees in dynamic task planning.
- Implemented an end-to-end pipeline on the V-REP CoppeliaSim simulation environment and on a real Fetch robot for table-top settings.
- Thesis available online.