About collecting data with my own robots

[YanJiaHuan]

Hi, Thanks for your sharing. Currently, i am reproducing your work with my own robot arm(which is 6-dof). And when i run the code: python -m furniture_bench.scripts.collect_data --furniture one_leg --out-data-path teleop_sim_demo --is-sim --scripted to collect data, I met some problem.

First of all, i already did the following:

1. I changed furniture_sim_env.py, config.py and sim_config.,py only to decrease the 7-dof setup to 6-dof (the change is brutal only to testify if i can run it).
2. I changed the urdf from franka panda to my robot+gripper.
3. Now the issac gym's rendering is correct (already changed to my robot).
4. I can run python -m furniture_bench.scripts.collect_data --furniture one_leg --out-data-path teleop_sim_demo --is-sim --scripted, and it seems the script is run correctly.

The target of mine:

• Use my robot to collect data, training, evaluation.
• Deploy it on real world.

The problem I met:

1. Since the current change is brutal, the codes can run, but the robot is not doing its work, with even a wired initial position.

The question or asking:

1. Can you provide some hints (instructions will be better) to tell us if we want to deploy your work on a new robot, what is something we need to do?
2. How do you modify your hype-para like 'reset_joints', 'position_limits'...Where can i get my correct numbers?

Hope to get some advice.