Walker S2 EDU: Real-Robot Deployment (Rollout)

Deployment script: ubt_IL/scripts/deploy/walker_s2/rollout.sh, shared by simulation and real-robot deployment (differences are described in the workflow pages). This page covers all parameters, safety prechecks, and online evaluation.

Prerequisites

• The robot is powered, standing, and in developer mode.
• Startup sequence: power on, enable servos and press D to start internal motion control -> place the robot on the ground, hold A to enter standing mode. See the Walker S2 EDU development guide for remote details.
• Walker real deployment runs on the robot Vision board; copy the project and model there, then build and start the container. The build script detects the board and builds the corresponding ARM container.

Deployment matrix

Scenario	Key command
19D simulation model -> 19D PGC gripper hardware	ROBOT_MODEL=walker_s2_19d
10D simulation model -> real deployment (act_async)	ROBOT_MODEL=walker_s2_10d INFERENCE_TYPE=act_async
Simulation feedback validation	ZMQ_HOST=127.0.0.1 (start the simulation task and bridge first)
Long-running deployment	Use the inference server (no cold start)

Deployment steps (real robot)

# 0. Enter developer mode -- ubt container on the robot control PC
ssh -p 2222 ubt@192.168.11.2    # password: Ubtubt@9880
# Request developer mode: true to enter, false to exit
ros2 service call /sys/task/developer_mode std_srvs/srv/SetBool "{data: true}"
# true = developer mode, false = normal mode
ros2 topic echo /sys/state/walker_mode

# 1. Build and start the inference container -- robot Vision board
# 1.1 Copy the project (code + models) to the Vision board
scp project-code-and-models /home/walker/
# 1.2 Log in to the Vision board and build the container
ssh walker@192.168.11.3          # password: aa
cd project-path/ubt_IL/docker
bash run.sh build
# 1.3 Start the container
bash run.sh start
bash run.sh bash

# 2. Initialize the pose (raise the arm above the table; use --home to return to zero)
bash /ubt_IL/scripts/deploy/walker_s2/robot_ready.sh

# 3. Deploy the real-robot 10D model with asynchronous inference
ROBOT_MODEL=walker_s2_10d \
POLICY_PATH=/ubt_IL/model/walker_pick_part_real_10d_2RGB_act/checkpoints/080000/pretrained_model \
INFERENCE_TYPE=act_async INFERENCE_HZ=1 \
FPS=13 DURATION=60 \
bash /ubt_IL/scripts/deploy/walker_s2/rollout.sh

# 4. (Optional) Verify the camera path; specify camera topic/robot config
/usr/bin/python3 scripts/deploy/walker_s2/preview_camera.py --robot walker_s2_10d

Know the two machines: - Jetson 192.168.11.3 (SSH walker, password aa): runs the lerobot-tienkung container and inference. - Robot control PC 192.168.11.2 (SSH ubt, port 2222): developer mode / ROS 2. It does not run the inference container.

Demo

Initialize / home (robot_ready.sh)

Command	Purpose
robot_ready.sh	Ready pose (--init, three safe stages, 13 s by default)
robot_ready.sh --home	Return to zero (three stages, 15 s by default); use after inference
robot_ready.sh --legacy-init / --legacy-home	Fall back to the old four-stage flow
Other arguments	Passed through to robot_control.py (e.g. --init-duration 20)

Safety precheck

Deployment runs an action dimension precheck: policy output_features.action.shape[0] must match robot config action_order; otherwise deployment is refused.

• ROBOT_MODEL must match the training DOF (see below).
• If the policy has no action names, set ALLOW_DIM_ONLY_POLICY=1 (enabled by default).

Environment variables

Variable	Default	Description
POLICY_PATH	Required, no default	Checkpoint directory (containing config.json)
ROBOT_MODEL	walker_s2_31d	ROBOT_MODELS registry key (see below)
ROBOT_CONFIG	configs/<ROBOT_MODEL>.json	Full path to the robot config (override for custom configs)
ALLOW_DIM_ONLY_POLICY	1	Allow dimension-only matching when the policy has no action names
STRATEGY	base	Rollout strategy
FPS	13	Control rate
DURATION	30	Run time (seconds)
TASK	walker s2 rollout	Task description
PREVIEW_CAMERA	1	Show the camera preview window
RECORD_ACTIONS	1	Record rollout actions (RECORD_OUTPUT_DIR sets the output)
INFERENCE_TYPE	sync	Inference engine (sync / act_async)
INFERENCE_HZ	0.6	act_async replanning rate
EXECUTION_HORIZON	0	Truncate chunks after N steps (0 = no truncation)
ZMQ_HOST	127.0.0.1	Bridge address (local for simulation feedback; robot IP for cross-machine deployment)

Exported parameters: BLEND_HORIZON=10, BODY_PUBLISH_HZ=300, BODY_V_MAX=2.0 (bridge chunk blending and publish rate limiting).

Robot configuration (ROBOT_MODELS registry)

Registry key	Dimension	End effector	Config file
walker_s2_10d	10D (right arm 7 + head 2 + right gripper 1)	PGC right gripper	walker_s2_gripper_19d.json (10D is a subset)
walker_s2_19d	19D (17 body + 2 grippers)	PGC 1-DOF gripper	walker_s2_gripper_19d.json
walker_s2_31d	31D (17 body + 14 hands)	V4 dexterous hand 7 DOF	walker_s2_v4_hand_31d.json

Online evaluation (recording inference trajectories)

RECORD_ACTIONS=1 (enabled by default) records rollout actions for analyzing model predictions, robot execution, and ACT blending. Records are saved under ubt_IL/scripts/deploy/output.

Offline policy evaluation

Run the offline MSE evaluation (ubt_IL/scripts/eval/eval_policy.py, shared with TienKung) before deployment; see Imitation Learning Platform §6.

Inference server (warm start)

For long-running or multi-task deployment, use the inference server to avoid a cold start (verified on hardware 2026-08-09: ~6 s warm-up to READY; start/stop/home/shutdown all pass). It supports start/stop/home/status/shutdown, model hot-swap with load, and remote launch. See Inference server.

start / home / shutdown move the robot; execute them next to the robot.

FAQ

Symptom	Fix
Deployment reports a dimension mismatch	Safety precheck refused: match ROBOT_MODEL to the policy dimension (10D -> walker_s2_10d, 19D -> walker_s2_19d, 31D -> walker_s2_31d)
Robot does not move	On hardware, confirm Bridge2 is running (5561/5562), the camera relay is running (5563), and developer mode is active; in simulation confirm ZMQ_HOST=127.0.0.1 and the simulation task is running
Inference server start fails	start/stop require INFERENCE_TYPE=act_async; the sync engine has no pause/resume semantics
launch --remote fails	Confirm --host is the Jetson 192.168.11.3 (SSH walker), not the control PC; the SSH user needs sudo docker exec permission
Repeated inference server launch fails	The singleton PID file /tmp/walker_inference_server.pid exists; shutdown cleanly or remove it manually
Simulation and hardware camera sizes differ	Simulation RGB is [3,240,320] (10D) or [3,480,640] (33D); hardware is [256,320] etc. Model input must match the training data — do not mix them