A SHORT HISTORY OF ROBOTS AND THINKING MACHINES

Although robots are considered a 20th-century invention, their origins lie in the distant past. From the earliest times, people have created myths about mechanical beings built in their own likeness with superhuman powers. The ancient Egyptians and Greeks built mechanical automations to perform simple tasks. In modern times, mechanical toys entertained and ever-more-complicated machines were invented. The idea of a lifelike mechanical humanoid influenced both art and science; in 1818, Mary Shelley's Frankenstein explored what happens when a man-made monster is given life by a well-meaning scientist. As computer technology became more advanced, scientists became more interested in building intelligent machines that could eventually think for themselves. Today, robots of all kinds populate our world and are used for varied applications in space exploration, the military, medicine, industry, research, police work and, of course, the movies. Here are some highlights of robot invention in the 20th century.

Calling a meeting and cleaning up a tennis court are not hard jobs- for people. But for robots? Contests at the annual meeting of the American Association of Artificial Intelligence (AAAI) in 1996 challenged robots to do just that. Competing robots had to navigate a maze of offices in one contest and locate and collect tennis balls in the other. Join FRONTIERS in the fun to see how the latest generation of autonomous robots complete the assignments almost as if they were alive!

The robot contest at the annual meeting of the American Association of Artificial Intelligence is designed to demonstrate the best efforts from the fields of artificial intelligence (AI) and robotics. Contestants must design a robot with enough intelligence and capabilities to participate in a challenge event like these: "Call a Meeting" required robots to navigate a maze of offices and corridors and summon a meeting at a specific time; "Clean Up a Tennis Court" challenged roots to find and sweep up tennis balls.

As humans, we might look at the events' descriptions and wonder why robots are asked to complete such seemingly easy jobs. We would outline a procedure (steps) necessary to accomplish the task, then do it. Defining the steps, or making an algorithm, is also one of the first tasks a robotics researcher would take. The algorithm lays the groundwork the researcher will use to develop a robot and controlling program that will complete the challenge.

As you see on FRONTIERS, defining a procedure, or algorithm, even for a seemingly simple task is much more difficult than it appears. A paradox of artificial intelligence is that the easier a task is for humans to perform, the harder it is to create a computer program to model it.

In the first activity below, you will design an algorithm and "program" a human " robot" to complete the task. For these activities, you will want to work in teams. To add a little fun to this activity, have the teams in your class compete against each other to see which can complete the event in the fastest time. The second activity is based on one of the events you'll see in this episode.

Understand a programmer's job by writing instructions for a robot to perform specific tasks.

Work in teams of three or four students each to define an algorithm, write a "program" for a "robot" to follow, have your robot execute the program to see how effective your algorithm is and "debug" your program to improve it. The team will collaborate on writing the program; then one of the members will act as the robot that executes it. You'll want waterproof aprons or coveralls or ever a change of clothes for this event.

Place two cups, one filled with water, the other empty, on a table in front of a chair. Your "robot" must sit in the chair and pour the cup of water into the empty cup. Oh, by the way, your robot will be blindfolded and won't be able to see.

1. Brainstorm. As a class, brainstorm what steps a robot must follow to pour water from one cup to another. List on the board as many steps as you can. Be specific.

2. Design an algorithm. Using the steps listed on the board, people on each team should write down, in order, the steps that will allow their robot to accomplish the challenge. (See example on p. 7.)

3. Write the program. Write down specific instructions your robot will carry out to complete each step. Your robot will follow these instructions as they are read by a team member, while another person or your teacher makes sure they are followed exactly.

4. Test your robot. Blindfold your robot and have it sit in a chair. One team member will read the program and the robot will execute the instructions. The other team members should takes notes on any problems that occur as your robot follows the instructions. (The team member who reads the instructions should also be careful that the robot does not accidentally get hurt.)

5. Revise the program. Now you must debug your program by revising the instructions your robot will follow. Make sure you include any changes the team noted when testing your robot.

6. Test the updated robot. Have members of the team exchange roles so the new program is run on a robot that has not "learned" from previous attempts.

Note that the sample algorithm and program on the next page, if executed exactly, should not work unless the robot "cheats"! For example, such things as which arm to lift and how high, how far forward to move the arm and how to determine if the robot has grabbed the cup containing the water are not defined. Question: How does the robot know which cup contains the water?

Other programs you can write for your robot might include:

*Making a peanut butter and jelly sandwich.

*Tying a shoe.

*Putting on a pair of boots.

*Opening an umbrella.

Driving without putting your hands on the steering wheel is not usually recommended, but today it's possible to ride in an experimental van that does the driving for you. For more than ten years, robotics engineer Chuck Thorpe of Carnegie Mellon University in Pittsburgh has been working on a project call Navlab, with the ultimate goal to design and produce a vehicle that drives itself. Using sensors and video cameras, the van navigates highways without a human driver.

A robot is a device made of two main parts: a mechanical or machine part, and a control device or brain. In simple terms, the Navlab vehicle also consists of these components: a machine (van) and computerized components that read the read and tell the can what to do.

Among the more positive applications of robotics are the many high-tech devices that help disabled people function normally. A few years ago, a design team at Carnegie Mellon University invented Pizzabot, a robotic arm that could make a pizza. With the help of this high-tech chef, a disabled person used voice commands to tell the robot which topping to put on the pizza.

Surprisingly, a pizza is a complicated product to make. Each step in the pizz-making process must be programmed. Using the algorithm procedure on pages 6-7 of this guide, describe the sequence of steps involved in making a simple pizza.

Now, try to build a pizza using just your voice. By using verbal commands, you will guide a mechanical device over a "pizza" target. The machine is a radio-controlled car with a marking pen attached to it. Since an electronic voice-activated circuit would be too difficult to build and too expensive to buy, we'll use the next best thing - a human voice. Work in teams of four.

Simulate programming a robotic device.

1. Assign each person in the group one of these roles: Engineer - figures out how to attach the marker to the R/C car. Timer - times each person's turn. Operator - directs the movements of the R/C car by voice command; can give only four commands: forward, stop, right, left. Controller - the only team member allowed to work the controls of the R/C car; responds only to commands issued by the operator.

2. Have the engineer attach the marker to the R/C car so that the marker's point extends beneath the car and makes contact with the floor surface. If attached properly, the marker will leave a trace of the car's path.

3. The timer should tape the pizza to the floor. The engineer should place the modified R/C car in the center of the newsprint circle, making sure the marker remains in contact with the newsprint surface.

4. The controller (person holding the control unit) should stand about ten feet from the target, facing in the opposite direction.

5. The operator, while facing the target, then gives commands to move the R/C car. The object is for the operator to have the car trace "cuts" onto the pizza target that will result in four equal slices.

6. The timer gives the operator three minutes to complete the objective. At the end of three minutes, team members should exchange places, making sure everyone has a chance to perform each of the four roles.