CS 5335: Robotics Science and Systems – Final Project General Information
The project offers an opportunity to apply concepts/techniques learned in class on a substantial problem that interests students. Students may work alone, in pairs, or in teams of three. We recommend students work in teams of two. Overall, we expect each team member to spend about 25–30 hours on the project (about 6-8 hours per week for four weeks). Therefore, a team of two must take on a more challenging project than a team of one. A three person project must be more challenging still.
Topics
The final project could be an application of an algorithm (either taught in this class or otherwise) to some robotics problem of interest to you. If you want, you can explore novel variations of an existing algorithm or develop a novel algorithm in its entirety. Some possible topic ideas are listed below, but feel free to select any reasonable topic that interests you.
• Develop an application that enables a robot to stack blocks or create some kind of structure. How do you perceive block positions and orientations? How should you plan manipulation behaviors? What assumptions do you need to make?
• Develop an autonomous driving system. How do you perceive the road and other relevant information? How will you plan or control the car’s movements?
• Compare different motion planning algorithms on a challenging problem. Explore non-trivial algorithm variations. Evaluate the effect of different techniques and identify what types of problems these techniques are most helpful for.
• Identify some problem in the OpenAI Gym https://gym.openai.com/ and solve it using some method.
Most projects should be evaluated on a robotics simulator of some type. Possible simulators you might consider using include: PyBullet (https://pybullet.org/wordpress/), Gazebo (http:// gazebosim.org/tutorials?tut=ros_overview), OpenRave (http://openrave.org/), or Mujoco (http://www.mujoco.org/).
Expectations
Your team is expected to find a problem of interest, formulate it well and precisely, develop/apply an algorithm to solve the problem, analyze your results, and communicate that to the class (and to the teaching staff).
Specifically, you must do the following:
• Identify a non-trivial problem that you plan to try to solve.
• Identify an algorithm (does not need to be novel) that you plan to implement/explore and a baseline method against which you will compare your method.
1
• Perform the comparison in simulation, using one of the simulators mentioned above (PyBullet, Gazebo, OpenRave, Mujoco). If you want to use a different simulator, that might be fine too, but make sure you discuss it in the project proposal.
• Obtain quantitative results from the simulation and describe them in your final report. Dis- cuss the results and explain why you think they make sense or why they don’t make sense. It’s okay to obtain weird results, but make an effort to understand them and perform experiments that could shed light on what is going wrong.
Deliverables and Timeline
Project proposal: You should submit a one-page project proposal that describes what problem you plan to tackle, what algorithms you plan to apply, what simulator you plan to use, and what experiments you plan to run.
Project milestone: Please submit a half-page report describing your project progress. By this date, we expect you to have installed the simulator and begun implementation of the algorithm. By this date, you should have identified any “show stoppers” that would necessitate major changes in project direction.
Final project report: This deadline is strict, because final grades are due soon after that, and we need time to grade the projects. The report should be approximately 6–8 pages (including figures) to include all of the above. You can write as much as you wish, but do not be excessive – just get to the point and provide a complete description of the project, including the components described above.
Project presentations (last one or two class sessions): Project presentations will happen in the final one or two class periods. These classes will be held entirely on Zoom from the students’ respective computers. Presentation slots (between three and five minutes each) will be assigned on a semi-random basis, after taking conflicts into account.
Grading
The project as a whole is worth 25% of your grade. That breaks down as follows: project proposal (3%), milestone report (1%), final Zoom presentation (3%), final project report (18%). The primary grading criterion will be thoroughness of the work and completeness of the results for a non-trivial project topic. We want to see
2