Event
PhySim Gate releases first version v0.1.0
1 reports · 1 sources
kiyeonjeon21 released the first version of PhySim Gate v0.1.0 on GitHub. It is a simulator-agnostic regression gate tool, applicable to simulation results for robotics, embodied intelligence, autonomous driving, and drones. Given a baseline run, a candidate run, and gate.yaml, the tool judges metric by metric whether the candidate has degraded and explains the reasons, outputs JSON, JUnit XML, and a standalone HTML report, and returns CI-usable exit codes (0 pass, 1 regression, 2 input error, 3 uncertain or invalid, 4 internal error). Running requires no GPU, simulator, or container. Currently the tool is not yet available on PyPI and must be installed from this tag or the accompanying wheel; adapters for LeRobot, CARLA, ROS 2, and PX4 have not yet been implemented.
Report timeline
- PhySim Gate v0.1.0 released: a regression gate tool for simulation results such as robotics
PhySim Gate v0.1.0 is the first release of a simulator-agnostic regression gate tool for robotics, embodied AI, autonomous driving, and drone simulation results. Given a baseline run, a candidate run, and gate.yaml, it judges metric by metric whether the candidate has degraded and explains why, outputs JSON, JUnit XML, and a standalone HTML report, and returns CI-usable exit codes (0 pass, 1 regression, 2 input error, 3 inconclusive or invalid, 4 internal error).
GitHub Embodied AI Topic ·