This is part of a series of posts that summarize my meetings with Professor Ian M. Mitchell for our own records.
During the third meeting, on February 2nd, we discussed the relevance of various control theory subfields, reviewed the scenario drafts, and defined areas of improvement.
This week’s highlights
- Completed weeks 1-4 of Georgia Tech’s Coursera course on the Control of Mobile Robots
- Further readings from Siegwart’s Intro to Autonomous Mobile Robots, specifically relating to the boundary value problem.
- Studies into differential equations, both ordinary and linear. Practiced finding exact solutions to the BVP as well as computing approximate solutions with finite difference approximations.
- Wrote a set of behaviour specific scenarios for evaluating sailboat controllers.
Summary of discussion
- My naturally evolving exploration of the field of autonomous robotics has resulted in a lacking understanding of control theory fundamentals. I spent the week bringing these up to par.
- The behaviour specific scenarios may still be useful for controller/planner evaluation and refinement but won’t be used in the final paper. Also note that the time required to implement the entire test suite may slow development.
- Evaluating simulation tests is generally done by checking the following criteria (in order of priority):
- Solution found (goal reached)
- Numeric score (if applicable)
- Visual confirmation (sometimes it is non-trivial to evaluate the solutions purely with a score)
Goals for next meeting
- (early goal) Prepare a brief reading list of the papers most relevant to the target problem.
- Explicitly define one or two key scenarios that will be used in evaluating the proposed method in my conference paper. This will also require defining the state space and differential constraints of the vessel.
- Ensure to define the resolution used for planning, distance to the goal, and the dynamic nature of obstacles.
- Create initial simulations for select controllers presented in the papers mentioned above.