|
PSO-based leg-loss identification method for legged robots |
|---|---|
| รหัสดีโอไอ | |
| Creator | S. CHATTUNYAKIT, Y. KOBAYASHI and T. EMARU |
| Title | PSO-based leg-loss identification method for legged robots |
| Publisher | Thai Society of Mechanical Engineers |
| Publication Year | 2559 |
| Journal Title | Journal of Research and Applications in Mechanical Engineering |
| Journal Vol. | 4 |
| Journal No. | 1 |
| Page no. | 46-57 |
| Keyword | Fault detection, self-identification, normalized cross-correlation |
| ISSN | 2229-2152 |
| Abstract | Legged robots have been widely developed and utilized in various applications since they are more flexible than the conventional wheel-based robots which cannot perform effectively in bumpy areas. Although the legged robot has no difficulty operating in uneven terrain, the broken parts of legged robot can lead to the task failure. In case of damages, the legged robots cannot operate properly with prior control strategies due to transformed models. This paper proposes the new method to detect the broken legs by employing only internal sensors. The lengths of robot legs will be estimated using the comparison between damaged robot and candidate models constructed in the simulation. Particle Swarm Optimization (PSO) is operated to discover the best candidate model that provides the highest fitness value. The similarity of orientation of robot body between actual damaged robot and candidate models is set as the fitness function calculated using normalized cross-correlation algorithm. The efficiency of this method is verified using numerical simulations and experiments, which shows that the proposed method can detect the lengths of robot legs more accurate than the existing method. |
