Biomimetic Gait Modeling from Small Walking Creatures in Nature Using Motion Compensator
Disciplines
Applied Mechanics | Artificial Intelligence and Robotics | Computer-Aided Engineering and Design | Dynamical Systems | Dynamics and Dynamical Systems | Engineering Mechanics
Abstract (300 words maximum)
There is a lot to learn from the world of insects. Their ability to continue moving despite losing limbs can help advance the future of robotics by inspiring the creation of adaptable robots capable of functioning even after losing legsThe main goal of our project is to develop a locomotion compensator for the position without controlling the orientation of the insect and to use machine learning software called DeepLabCut (DLC) to track its gait pattern. This data can be utilized for future applications, such as in robotics. The developed system allows insects to move freely without restraint, enabling us to obtain accurate gait patterns from the motion modeling software. Using past projects with a smaller locomotion compensator, we reverse-engineer and redesign the system to create a larger, more robust version that can support greater mass and track data for larger insects. The previous design used three omni-wheels, whereas our system employs two omni-wheels and a trackball for position control. The trackball provides additional sensory information, which we use to move the motor more accurately in response to data from a camera mounted on top of the system. We use DLC to accurately capture the insect’s movement from a video of it moving on the sphere. We labeled body parts of the insect in images to create a training dataset. The machine learning model is then trained on this dataset and used to evaluate the entire video to identify the gait patterns. With these advancements, we aim to develop more robust and adaptive robotic systems and enhance the understanding of insect locomotion.
Academic department under which the project should be listed
SPCEET - Mechanical Engineering
Primary Investigator (PI) Name
Dal Hyung Kim
Biomimetic Gait Modeling from Small Walking Creatures in Nature Using Motion Compensator
There is a lot to learn from the world of insects. Their ability to continue moving despite losing limbs can help advance the future of robotics by inspiring the creation of adaptable robots capable of functioning even after losing legsThe main goal of our project is to develop a locomotion compensator for the position without controlling the orientation of the insect and to use machine learning software called DeepLabCut (DLC) to track its gait pattern. This data can be utilized for future applications, such as in robotics. The developed system allows insects to move freely without restraint, enabling us to obtain accurate gait patterns from the motion modeling software. Using past projects with a smaller locomotion compensator, we reverse-engineer and redesign the system to create a larger, more robust version that can support greater mass and track data for larger insects. The previous design used three omni-wheels, whereas our system employs two omni-wheels and a trackball for position control. The trackball provides additional sensory information, which we use to move the motor more accurately in response to data from a camera mounted on top of the system. We use DLC to accurately capture the insect’s movement from a video of it moving on the sphere. We labeled body parts of the insect in images to create a training dataset. The machine learning model is then trained on this dataset and used to evaluate the entire video to identify the gait patterns. With these advancements, we aim to develop more robust and adaptive robotic systems and enhance the understanding of insect locomotion.