Abstract | ||
---|---|---|
Sensing visual motion gives a creature valuableinformation about its interactions with the environment. Flies inparticular use visual motion information to navigate through turbulentair, avoid obstacles, and land safely. Mobile robots are idealcandidates for using this sensory modality to enhance theirperformance, but so far have been limited by the computational expenseof processing video. Also, the complex structure of natural visualscenes poses an algorithmic challenge for extracting usefulinformation in a robust manner. We address both issues by creating asmall, low-power visual sensor with integrated analog parallelprocessing to extract motion in real-time. Because our architectureis based on biological motion detectors, we gain the advantages ofthis highly evolved system: A design that robustly and continuouslyextracts relevant information from its visual environment. We showthat this sensor is suitable for use in the real world, anddemonstrate its ability to compensate for an imperfect motor system inthe control of an autonomous robot. The sensor attenuates open-looprotation by a factor of 31 with less than 1 mW power dissipation. |
Year | DOI | Venue |
---|---|---|
1999 | 10.1023/A:1008916202887 | Auton. Robots |
Keywords | Field | DocType |
optomotor system,analog VLSI,optic flow,motion detector,insect vision | Computer vision,Simulation,Computer science,Biological motion,Visual sensor network,Motion detector,Artificial intelligence,Autonomous robot,Very-large-scale integration,Stimulus modality,Detector,Mobile robot | Journal |
Volume | Issue | ISSN |
7 | 3 | 1573-7527 |
Citations | PageRank | References |
17 | 1.76 | 9 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Reid R Harrison | 1 | 222 | 57.49 |
Christof Koch | 2 | 7248 | 973.47 |