Results/Conclusion
Our wheelchair model successful detects the van ramp and plans a path to ascend the wheelchair. Our simplified controller works well in simulation and correctly gives torque inputs to drive up the ramp.
Difficulties
With unreal engine, we ran into multiple problems with rendering the graphics, running the simulation due to missing Big Sur packages, and unreal engine physics. Due to these problems, we found we could only run unreal on Windows computers. Moreover, we had slow rendering times with computers without a gpu. Additionally, our simplified controller worked well in simulation with the artificial physics but real world physics may vary from simulation. We had a little trouble integrating all the pieces of our project together as we were each in charge of an individual model. There were many middle boundaries that crossed multiple boundaries that we had to cover.
Potential Improvements
If we had more time, we would have liked to implement point cloud navigation in which we have a server that remembers the point cloud for future wheelchair navigation. This map of the environment would ideally benefit in house mobility as the wheelchair would actively be able to path plan out of room. Additionally, we would have liked to tune our controller and make it less jerky as well as implement a improved path following algorithm.