vo_benchmark

bench mark codes for VO, truncated VO, RVO and ORCA Side-by-side benchmark of VO / τ-VO / RVO / ORCA collision avoidance algorithms, visualised in a 2×2 animated plot.

Install

pip install numpy scipy matplotlib pyyaml
pip install osqp        # optional but recommended (faster QP solver for ORCA)
sudo apt install ffmpeg # optional, required for .mp4 export

Usage

python3 run.py                              # default: circular, 4 agents
python3 run.py --scenario frp --n 6 --seed 42
python3 run.py --scenario headon
python3 run.py --save result.gif
python3 run.py --save result.mp4            # requires ffmpeg

Settings are loaded from config.yaml automatically. CLI args override it.

config.yaml

scenario:    circular   # circular | frp | headon
n_agents:    4
radius:      5.0        # initial circle radius [m]
v_pref:      1.0        # preferred speed [m/s]
r:           0.5        # agent collision radius [m]
tau:         5.0        # time horizon [s]
dt:          0.05       # timestep [s]
n_steps:     400
n_samples:   40         # velocity candidate grid size for VO/τ-VO/RVO
interval_ms: 60         # animation frame interval [ms]
cone_scale:  0.9        # forbidden region overlay scale
seed:        0