Robotics Pioneers and the Machines They Helped Build

Giving a factory arm a stored sequence of moves Devol’s programmable arm could follow commands stored for it in advance. The idea reached factory work through Unimation. Start here to see why a machine designed to do a job repeatedly could still be valuable when the job changed: a new sequence gave the same arm different work.

Using computers to control an electric arm Scheinman’s Stanford Arm put electrical power and computer control together in a machine that researchers used to explore assembly. Put it alongside Devol’s factory arm to see the difference between proving a programmable machine useful on a line and giving researchers a way to test new manipulation ideas.

Putting perception and planning on wheels Rosen led the SRI group behind Shakey, a mobile robot that could use information from its sensors to plan how to complete a task. Its camera and touch sensors sent information to a separate computer. The separation makes it easier to see the job the system had to solve: observe a room, decide what to do in it, and then move.

Working out where a robot should go next Nilsson worked on planning for Shakey and helped develop a way to search for routes with colleagues Peter Hart and Bert Raphael. Read him alongside Rosen for a closer look at one part of the same project. Knowing where something is gives a mobile robot a problem to solve, and a planner has to turn that information into a route.

Building a gripper that could adapt to a shape Hirose drew on work with snake-like robots to develop a flexible hand that could hold objects gently. The Soft Gripper shows why the shape of a machine matters as much as its controls. When the hand can conform around what it touches, there are more ways to grasp a fragile object than closing two stiff fingers against it.

Getting a cart across a cluttered room Moravec used a sliding camera to give the Stanford Cart several views of obstacles. In 1979 the cart made its way across a room without a person steering it, but the trip took hours. That combination makes it a good example of the gap between showing that a machine can get somewhere and making it quick enough for everyday use.

Finding corresponding details in images Kanade and Bruce Lucas developed a way to match details across images. Their 1981 paper explains how the technique could help with stereo vision and motion analysis. The paper is a good place to see what computer vision contributed to robotics. Recognizing that two images show the same feature is a step toward telling how far away it is or how it moved.

Inspecting places people could not safely enter Whittaker and colleagues developed robots to inspect the aftermath of the Three Mile Island accident. The project is an early example of field robotics. A machine built for the site had to be useful in a real environment it could not change, a different test from getting a machine to move in a laboratory.

Taking a small rover to Mars Shirley oversaw development of the microrover experiment for Mars Pathfinder. Sojourner, the six-wheeled rover, reached the Martian surface in 1997. Its small size helps explain the problem her team faced. A machine that had to travel with a lander still needed to move around and carry instruments once it got there.

Helping a surgeon hold a tool steady Taylor and colleagues developed a steady-hand system in which the surgeon and robot hold a tool together. The machine responds to the surgeon’s movements while helping with steadiness. It gives you a concrete example of medical robotics in which the machine is designed around how a person works.

Giving a robot cues for taking turns Breazeal’s Kismet used cues such as vocal timing and expression to manage a simple back-and-forth with people. Look for the pauses and changes of attention in a demonstration. The robot’s job was to make an interaction legible enough for a person to know when to respond.

Bringing home robots out of the lab Greiner co-founded iRobot with Colin Angle and Rodney Brooks and helped take the Roomba from development to launch. The vacuum gives you a way to think about a different engineering test. A robot that can move around a room has to do a job people recognize as useful and fit into homes that were never designed for it.

Bringing warehouse shelves to the people filling orders D’Andrea, Mick Mountz, and Peter Wurman built Kiva’s system of mobile robots that move racks of stock. The useful change was where the walking happened: the shelves came to the workers. Read this entry to see how a group of relatively simple machines can change how a whole workplace is laid out.

Giving a remote operator a sense of touch underwater Khatib led the team behind OceanOne, a robot with two arms and touch feedback for its operator. The feedback matters when the work is to handle something fragile at a depth a diver may struggle to reach. The project shows how robotics can extend a person’s judgment into a place where they cannot easily work themselves.