Android Bodies

In another post I blogged about how a humanoid robot (sometimes called an android) could be built. I divided the task into three parts, the robot's body, brain (or computer) and its programming (intelligence). In this article I'd like to get into more detail about what is required to design and build the android's body.

First it needs a power supply. At present this is likely to be chargeable batteries. Other possible sources could be compressed gases, hydraulics, flywheel energy, the decay of organic material, nuclear fusion, when and if it ever becomes available, or other radioactive sources, solar energy, etc. For the purpose of a man-like frame, many of these possibilities are either too bulky or too complicated.

The next thing we need to think about are actuators which are the parts that convert the stored energy into movements. In humans and animals it is muscles that do this job. At present most robotic actuators are electric motors which turn a wheel at a joint. A spring can be part of the motor actuator for improved force control, particularly needed for walking. Another method is to use wire that contracts and expands when electricity is applied and released. This would work similarly to muscles in a human being. New plastic materials that expand and contract have been used in facial muscles and arms of animatronic (humanoid appearing) robots.

One big problem to be solved is a sense of touch. A humanoid robot must be capable of determining how much pressure to apply with its hands to grasp items properly. Current robotic and prosthetic hands receive far less tactile information than the human hand. Recent research has developed a tactile sensor array that mimics the mechanical properties and touch receptors of human fingertips.

Another important sense is vision. Our android must be capable of recognizing and distinguishing between objects and of estimating their relationship to each other. There is an entire subfield of artificial intelligence concerned with designing systems that mimic the processing and behavior of biological systems.

Another difficult problem to solve is walking. Several robots have been made which can walk reliably on two legs. However, none have ever been built that are as robust as a human being. Some have said that Sony's robot Asimov walks as though it had to use the toilet. Nonetheless, several robots built by Marc Albert of MIT have successfully demonstrated very dynamic walking, even running and performing somersaults.

Speech recognition is a requirement if the humanoid robot is going to interact with human beings. Interpreting a continuous flow of sounds coming from a human in real time is a difficult task. Sometimes we don't understand each other when one has a different accent than he or she normally is used to. Currently the best systems can recognize continuous natural speech up to 160 words per minute with an accuracy of ninety-five percent.

There are many other design considerations. The ones I have listed are the most difficult to achieve.