Most efficient Inverse Kinematics algorithm for Quadruped models : Comparing FABRIK to CCD

Detta är en Master-uppsats från KTH/Skolan för elektroteknik och datavetenskap (EECS)

Sammanfattning: This paper compares the two heuristic inverse kinematics methods: Forward And Backward Reaching Inverse Kinematics (FABRIK) and Cyclic Coordinate Descent (CCD) in the use cases concerning quadruped models. Unoptimised versions of the two algorithms were implemented into a game engine and evaluated on a quadruped model. The two algorithms were evaluated by computational time, iterations and average error. The results show that FABRIK outperformed CCD in all of our test scenarios, in number of iterations required, average error as well as function execution time. However, results also showed that FABRIK was less superior when targets were barely within reach, since one leg might be able to reach a target, but two connected legs might not be able to reach two targets because of their interference with each other. This suggests that to improve on FABRIK it should be optimised when considering a bipedal or quadruped model. 

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)