dsh-ros2

ROS2 debugging toolset and robot-state vision analysis for DeepSeek Harness (DSH), shipped as a plugin. 中文版见 README_CN.md.

dsh-ros2 gives a DSH agent full robot development / debugging capabilities on any host with ROS2, organized in four capability tiers:

Tier	Capability	Safety boundary
L1	Read-only diagnostics: package/workspace/dependency checks, node/topic/service/action/param/interface enumeration, one-shot topic sampling, TF tree queries, whole-graph topology JSON, ros2doctor, bag summaries, MoveIt discovery, robot-profile load, safety-state read & VLM semantic arbitration	Pure read-only, no approval
L2	Approval-gated management: colcon build (background job), rosdep install, message skeleton generation, param set, bounded bag record, one-click ROS2 install, launch management, rosbag replay, MoveIt motion, zero-pose calibration, robot registration & topology learning, safety_monitor start / human-gated lock / unlock	Writes always ask first (fail-closed)
L3	Visualization: RViz2 / rqt lifecycle management, screenshots, multimodal vision description, xdotool-level window interaction	Local session operations
L4	Realtime vision: parallel VLM ROS2 node + image-topic acquisition (headless; vision_bringup periodically refreshes per-topic bridges), plus RViz2 offscreen rendering (OGRE kernel → /rviz/scene topic)	Pure software/GPU rendering, no display needed

All tools run plain ros2 / colcon / rosdep CLI commands on the host; L1 never modifies anything, L2 always asks first.

---

Screenshots

RViz2 offscreen render (latest lite_urdf, real material colors)	Head camera	Wrist-left camera	Wrist-right camera

Left: rviz_offscreen_node renders with the real rviz stack (OGRE) and publishes to the /rviz/scene image topic. Right: three frames grabbed from live camera topics by ros2_image_snapshot (1280×720). Full test record: docs/test-robot-state-vision.md.

---

Features

…