To supply the areas of conflict without putting anyone in danger, the Laboratory Computer Science and Artificial Intelligence at MIT runs a program to develop a forklift truck without a driver can handle the load and transport equipment almost independently. To achieve this, researchers are developing software related to networks of sensors and cameras allow the system to understand what surrounds it and react accordingly. “In Iraq, for example, unloading equipment must often be interrupted,” notes Matt Walter, one of the project. “It is possible to automate a number of tasks and thus reduce the exposure of people on site at risk.”
Understand and feel the space
Eventually, this robotic system can detect in its environment and the mere incitement or oral gestures of a person responsible for overseeing the distance. It may also avoid unexpected obstacles, such as the passage of human beings or falling objects. This, independently and faster than when it is under human control. For this it is necessary to reproduce many sensations like touch, movement, memory and deduction. “All this allows to represent a logic model of space and objects that compose it, and thus move with success,” recalls Seth Teller, professor of computer science. “But if people are able to build this mentally without thinking, this is not the case for the machine.
Recognition of the environment by laser
For now, a supervisor must explain precisely what the truck movements should be performed, and indicate on a touch screen to move the pallets. The recognition of people and obstacles and the ability to avoid, however, should be faster. “It is possible to detect movement of individuals using laser scanners,” notes Matt Walter, one of the project. “These rays are not really effective in identifying the presence of people who seek to hide or move very quickly.” The project involves about thirty researchers from institutions like MIT or BAE Systems. It was founded by the Agency U.S. Army Logistics Innovation.