🚀 Getting Started with Yumi

Calibration & Safety Notes (RAIL Lab)

Calibration

YuMi uses incremental encoders, so joint zeroing must be done manually:

1. Move arms to the marker-aligned calibration pose
2. On teach pendant → Calibration → Update Revolution Counters
3. Confirm for each arm

Auto Mode

• Auto mode runs ~10× faster
• Only enable when:
  • Workspace is clear
  • Program is verified in Manual mode
  • Calibration is complete

Networking Setup

Setting	Value
Service Port	XP23
Typical Robot IP	192.168.125.1
Host PC IP (example)	192.168.125.82/24

---

🎮 Control Interfaces

YuMi can be controlled through multiple software interfaces depending on the task, environment, and toolchain. Below are the recommended methods used in the RAIL Lab.

---

RAPID (On-Robot Programming)**

ABB's native robot programming language.

• Used for: production-reliable programs, onboard logic, dual-arm coordination.
• Edited in: RobotStudio (PC) or the teach pendant.
• Execution speed: Full performance, supports Auto Mode.
• Good for: stable manipulation tasks once development is finalized.

Reference: https://library.abb.com/ → Search “RAPID Instructions, Functions and Data Types Manual”

---

Python Control via Yumipy (RAIL Lab Primary Workflow)

• Repo (use this fork): https://github.com/galou/yumipy/tree/robotware6_06
• Purpose: Direct, script-level control of YuMi without ROS.
• Communication: TCP socket → requires controller modules:
  • SERVER_LEFT.mod
  • SERVER_RIGHT.mod
• Programming Style:

  from yumi import YumiRobot
  with YumiRobot() as yumi:
      yumi.left.goto_state([0, -130, 30, 0, 40, 0, 135], v=20, a=50)

Movement Command API:
https://berkeleyautomation.github.io/yumipy/api/robot.html

ROS Control + MoveIt

• Tested with Ubuntu 16.04 + ROS Kinetic
• Primary package: https://github.com/kth-ros-pkg/yumi
• Provides:
  • URDF model
  • MoveIt planning pipelines
  • Joint & trajectory controllers
  • Dual arm coordination

ROS Industrial ABB driver:
https://github.com/ros-industrial/abb

---