🧪 Tests to Ensure Reliable Motion Planning

[Victorlouisdg]

To deliver on our promise of plug-and-play motion planning for common use cases, we need a robust testing framework. This will empower us to consistently validate our solutions and easily benchmark different planner implementations (even those still under development).

Key Test Suite Requirements

• 🗺️ Diverse and Representative: Cover a range of real-world challenges inspired by the use cases at our lab.
• 🎯 Targeted: Isolate specific aspects and highlight planner strengths and weaknesses.
• 💡 Easy to Grasp: Named, simple, and clearly defined scenarios.
• 📈 Gradual Difficulty: Systematically increasing scenario difficulty pinpoints problems
• 🧐 Catch User Error: Discover problems in requested planning queries

To ensure systematic testing, let's organize our test cases in tables like this:

Joint goals 🦾
Check whether planning to goal joint configurations works. This can help isolate problems because inverse kinematics is not necessary for this.

Name	Purpose
01_no_joint_movement	🥱 Plan to the same joints as the start and goal
02_single_joint_movement	➡️ Plan to close goal joints where only one joint is rotated
03_linear_joint_movement	📏 Plan to far joints, but are collision-free when linearly interpolating
04_self_collision_joint_movement	💥 Plan to joints that would lead to a self-collision when linearly interpolating
05_env_collision_joint_movement	🚫 Plan to joints that would lead to an environment collision

User error handling 🧐
Check whether the planner correctly detects invalid user requests and terminates for infeasible plans.

Name	Problem
01_start_joints_invalid	⚠️ Attempt to plan from invalid start joints
02_goal_joints_invalid	⚠️ Attempt to plan to invalid goal joints