Control and hardware design for a bipedal robot via planar modeling with empirical perception studies leveraging embodied movement analysis

Robotics, Control, Bipedal Walking, Expressive Robots, Walking Robots

Huzaifa, Muhammad Umer

关键词: Robotics, Control, Bipedal Walking, Expressive Robots, Walking Robots;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/106157/HUZAIFA-DISSERTATION-2019.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Humans can walk across a range of surfaces without falling and can also communicate internal state to other humans through variable gait styles. That is why, developing bipedal robots is of interest for applications in human-facing settings. For the purpose of designing an expressive bipedal robot, this dissertation focuses on taking inspiration from the experiential understanding about human walking from the Basic Six in Bartenieff Fundamentals. To generate a range of stylistic walking behaviors, simplified planar biped models are studied under model-based trajectory optimization with variable constraints. From these walking behaviors, a set of gaits is identified and labeled, using embodied movement analysis, with stylistic verbs related to human walking, e.g., ``lope'' and ``saunter''. These labels are then validated by conducting user studies in Amazon Mechanical Turk and demonstrate that the gaits generated using our method are visually distinguishable and correlate with the human activity. These gaits are also investigated for the affect they can induce on the humans in the presence of visual stimuli in the form of background images. To implement the variable gait styles in real world, a hardware mechanism has also been developed that can replicate the notion of pelvic shift from the Basic Six. Thus, this dissertation lays groundwork for designing bipedal walking robots that can convey social cues with their movements and integrate harmoniously around humans.