Robotic Cable Insertion

Simeon Adebola, Corey Koehler, Rahul Shah, Kishore Srinivas, Karim El-Refai

Project Sponsor: Ken Goldberg

EECS 106A/206A Final Project, Fall 2022

Introduction

The goal of this research project is to sense and detect the position and orientation of a cable (rope or tubing) and insert it snugly into a channel.

Why is this interesting?

This task is something humans can do with ease, but doing it autonomously is difficult. The robot not only must exercise precision and care to manipulate the relatively thin cable, but also must be reliable in determining the location of each end of the rope, distinguishing the rope from the channel, and noting when the cable is thoroughly packed into the channel.

We start with our base goal, in which one end of the cable is attached to the channel, and the rest of the cable and channel are placed freely in the workspace. Our high level approach is to:

Use a depth sensor to capture an image of the workspace.
Identify the free end of the cable and the endpoint of the channel.
Place the end of the cable onto the end of the channel.
Press down along the length of the cable to fit the cable tightly into the channel.

20221209_130426.mp4

Notice how easy it is for a human to perform this task. Doing it autonomously is a far more involved process.

Identifying the position and orientation of the cable and channel endpoints from depth and grayscale images is the most challenging problem in this project, requiring precise thresholding and graph search algorithms.

Once we accomplish this base goal, we move to our stretch goal, where neither end of the cable is fixed to the channel. We take the same high level approach, but now need to identify both ends of the cable and channel and place both ends of the cable onto the channel. However, this task is significantly more complex since the cable has more freedom to move around and we cannot use the channel as an anchor point for moving our cable.

Applications

Our work in this project provides a foundation for applying robotics to real-world industry tasks such as routing cables in industrial settings, placing tubes on bicycle tires, and aligning cables before spooling to prevent messy spools. Automating these tasks, which are often done manually, will greatly increase throughput and lower costs, greatly benefiting factories and warehouses. Unfortunately, this may mean that those currently employed in these jobs get displaced by robots, as is the case in many other industries where automation has been replacing human labor. This is an important cost to keep in mind.

Routing industrial cables

Placing tubes on bicycle tires

Aligning cables to prevent messy spools