Sensors and Sensing for Reactive Robots Alexander Stoytchev Mobile Robot Lab Georgia Tech
Sensing: Conceptual Level
Sensor Classification Proprioception relative to internal frame of reference Exteroception measurements of the environment relative to robot’s frame of reference Exproprioception Measurement of the position of robot body or parts relative to the layout of the environment
Proprioceptive Sensors Shaft Encoders Internal Navigation System (INS) Global Positioning System (GPS) Regular Differential GPS (stationary base + moving robot)
Proximity Sensors Sonar IR Bumpers Radar Laser Range Finders Time of Flight Phase-based
Other Sensors Compass Accelerometers Inclinometers Gyroscopes Thermometers Microphones Cameras
Sensor Properties Field of view (FOV) Update rate Accuracy/Repeatability/Resolution Applicability to target domain Power consumption Hardware reliability Size Computational requirements Interpretation reliability
Desired Sensor Properties Simple to operate and maintain Modular Redundant (physical & logical redundancy) Fault Toloerant Cheap!
Sensor Errors False Positive False Negative Sometimes not clear Cost of Errors
Case Study: Sonar Calculate distance based on the time of flight d = ½ c t c = c0 + 0.6 T c0 = 331 m/s T- temperature in degrees Celsius Frequency – 50 KHz
Sonar Cone Regions
Problems with Sonars
Sensing: Conceptual Level
Raw v.s. Logical Sensors Raw Sensor Data Logical Sensor Data List of distances to obstacles Logical Sensor Data Egocentric polar/Carthesian coordinates Independent of physical sensor
Passive v.s Active Sensors Passive – the environment provides the medium for observation Active – the sensor puts out an energy into the environment != Active Sensing - use an effector to better position a sensor for observation
Sensor Fusion
Sensor Fission
Sensor Fashion