Infrared People Sensors For Mobile Robots

摘要

The MIT Mobile Robot Group is developing a new small, low power, autonomous robot which will interact with humans in dynamically changing environments. This paper describes a new sensor system based on a collection of pyroelectric elements which will enable the robot to initiate behaviors that include approaching and following humans. Pyroelectric detectors were chosen because they offer the capability of filtering out human infrared signatures from ambient surroundings, while also providing the advantages of simplicity, small package size, low power consumption and low cost. However, pyroelectric elements require relative motion between the sensor and the human to allow detection. In addition, while pyroelec-tric sensors are able to signal whether or not a target has moved through the field of view, they have no means of determining range or orientation to the targets, capabilities that would be desirable for a mobile robot. This paper describes a sensor configuration which gets around these problems and provides a low power means for a robot to ascertain distance and direction to people in its surroundings. The sensor system provides several information output channels which feed into the control system on different levels, triggering a variety of behaviors. Fixed sensors on the robot are used to detect multiple people or to ascertain the direction of a moving person. Rotating crossed sensors use geometric relationships to determine range and orientation to stationary targets. Using a behavior-based subsumption architecture control system, the sensors and information processing modules are organized so that the required behaviors can be produced without recourse to sensor fusion, enabling very low computational overhead.