Exoskeleton

I designed the overall architecture and the control system of a wearable robotic device that aims to enhance a human’s strength, carrying capacity and endurance.

Job-to-be-done

Design and build a self-powered exoskeleton including the overall architecture and control system.

my Role

Co-inventor and engineer

Results

2 commercialized patents and 1 spin off (Ekso Bionics)

About

The exoskeleton consists of two powered legs and backpack frame on which the power unit and an additional load can be mounted. The device is connected rigidly at the foot and more compliantly elsewhere. The exoskeleton allows a person to walk, turn and perform other movements like sitting and squatting. The device is controlled in a way that the human never really feels it. Using sensors and its computing power it captures the human’s intent and moves in the same direction. The human does not feel either the weight of the device or the load it carries on its back. The exoskeleton was developed at the Berkeley Robotics and Human Engineering Lab and then a Ekso Bionics.

The key challenges that we needed to solve in order to create a working prototype were:

How can the device sense where the human operator intends to move if the human can’t move separately from the machine?
How do we design a device that is strapped to a human and yet does not impede its movements or cause discomfort?
How do we design a device that can be used by humans of different shapes and sizes who move in different ways?

Process

I led the design of the overall architecture of the exoskeleton, meaning the logic of how it works and the layout of the limbs, joints, power systems, and sensors. My job was then to develop the robot’s control system (i.e. the nervous system and brains) to help it sense and shadow the operator’s movements. The process involved iterations around lots of computer simulations and lots of prototypes.

Lessons

I was able to develop a new robotic control method called “virtual force control” that allowed the exoskeleton to deduce the human’s intentions to move without using cumbersome and uncomfortable contact sensors between the human and the machine
Even though walking looks like a smooth continuous motion, the human body effectively behaves like a very different machine depending on how many feet and what parts of the feet are on the ground. This requires writing completely different sets of code guided by different equations for different movements..
Plan for the worst, when testing the machine. Expect things to break and expect the machine to move in unpredictable directions. Design the hardware and software accordingly.

Outcomes

In the end we were able to make an exoskeleton that walked both in the lab and outside the lab. It was selected in the New York Times Year of Ideas. And that was just the beginning. A company called Ekso Bionics spun out of our lab using our technology and patents. After several rounds of venture investments and steady growth, the company is now the global leader in robotic exoskeletons.