Project Details
The robot uses two mirrored legs with 2 DOF in the hips (ab/adduction and flex/extension), 1 DOF in the knee (flex/extension), and 2 DOF in the ankles (in/eversion and plantar/dorsiflexion). I tried to keep all the motors for the actuators as close to the hips as possible, but I did opt to stack the ankle hydraulic actuators on the inner thighs, rather than trying to put them on the pelvis area since I was hoping it would theoretically help to keep the center of gravity over the feet while the knees are at "rest" at 148 degrees.
Starting from the top and working down at the hip: you can see the hips use DCBL motors that drives the ab/adduction motion. The design for these DCBL motors is based on the design from Benjamin G. Katz at MIT in his thesis "A Low Cost Modular Actuator for Dynamic Robots". The ab/adduction motor is then secured to the flex/extension motor using an aluminum bracket, which is bolted in using M3 bolts (all fasteners on this robot are either M3 or M1.6 bolts, with theoretically all holes being tapped to mitigate the need for added weight in lock nuts. All bearings and attached axles would require an arbor press to assemble). Then, the flex/extension motor bolts to the knee motor using another milled aluminum bracket.
The knee uses a "spindle" which is bolted directly to the outer-most DCBL motor's rotating platform, while the structure of the thigh link is bolted to the stationary portion of the motor. The mechanism of the knee gave me a lot of trouble. In Katz's paper, he uses a belt and gear-driven system, similar in principle to how mine is designed, but with serious drawbacks in rigidity and stiffness. To combat this issue, I opted for a dual-direction cable design inspired by the LIMS2-AMBIDEX elbow joint from CMU.
Project Details
The robot uses two mirrored legs with 2 DOF in the hips (ab/adduction and flex/extension), 1 DOF in the knee (flex/extension), and 2 DOF in the ankles (in/eversion and plantar/dorsiflexion). I tried to keep all the motors for the actuators as close to the hips as possible, but I did opt to stack the ankle hydraulic actuators on the inner thighs, rather than trying to put them on the pelvis area since I was hoping it would theoretically help to keep the center of gravity over the feet while the knees are at "rest" at 148 degrees.
Starting from the top and working down at the hip: you can see the hips use DCBL motors that drives the ab/adduction motion. The design for these DCBL motors is based on the design from Benjamin G. Katz at MIT in his thesis "A Low Cost Modular Actuator for Dynamic Robots". The ab/adduction motor is then secured to the flex/extension motor using an aluminum bracket, which is bolted in using M3 bolts (all fasteners on this robot are either M3 or M1.6 bolts, with theoretically all holes being tapped to mitigate the need for added weight in lock nuts. All bearings and attached axles would require an arbor press to assemble). Then, the flex/extension motor bolts to the knee motor using another milled aluminum bracket.
The knee uses a "spindle" which is bolted directly to the outer-most DCBL motor's rotating platform, while the structure of the thigh link is bolted to the stationary portion of the motor. The mechanism of the knee gave me a lot of trouble. In Katz's paper, he uses a belt and gear-driven system, similar in principle to how mine is designed, but with serious drawbacks in rigidity and stiffness. To combat this issue, I opted for a dual-direction cable design inspired by the LIMS2-AMBIDEX elbow joint from CMU.
No comments:
Post a Comment