Goal: Fast (~30Hz) perception pipline for workspace approximation research for heuristic-based multi-agent planning framework (PMAF) by Laha et al 2023. Support both sim/real-world perception for multiple cameras.
| Hardware Stack |
|---|
| 1. CUDA-capable PC (Our Setup: i7 8000U + RTX3070) |
| 2. Intel Realsense 435i (for Real-World) |
| 3. Franka Emika Panda Robot Arms (for Real-World) |
| Software Stack |
|---|
| 1. Ubuntu 20.04 Focal Fossa (iso) |
| 2. ROS 1 Noetic (debian) |
| 3. Conda + CUDA (instructions) |
| 4. librealsenseSDK (debian) |
| 5. CoppeliaSim 4.1.0 EDU (for simulation,binaries) |
- Install aforementioned requirements and clone this git repo
- Run
./setup.sh- this will create a virtual environment and install all dependencies and libraries - Set
$COPPELIASIM_ROOTto the CoppeliaSim installation directory (for simulation)
- Riddhiman Laha
- Tinayu Ren
- CuPoch
- Open3D
- CoppeliaSim
- PyRep (by stepjam)
Performance w/o RBS:
