Discipline: Technology and Engineering
Subcategory: Civil/Mechanical/Manufacturing Engineering
Session: 2
Room: Exhibit Hall A
Danielle Carr - Florida State University
Co-Author(s): Carl A. Moore Jr., PhD, FAMU-FSU College of Engineering, Tallahassee, FL; Twan Capehart, Florida A&M University, Tallahassee, FL
In legged locomotion, robotic actuators are required to change operating regimes frequently while switching between low speed, high torque requirements and highly dynamic tasks. With large requirements, actuators are commonly operating outside of their efficient range. How to keep theses actuators close to optimal efficiency while maintaining the ability to achieve both tasks is the question. Verstraten et al. studied the use of two kinematically redundant actuators to meet the range of torque requirements necessary for locomotion and other applications (Verstraten 2019). Alternatively, Sugar and Holgate suggested that the use of a variable gear is a potential solution to achieving various robotic tasks with different energy requirements (Sugar and Holgate, 2013). Variable transmissions have proven their potential to reduce energy consumption of cars (Carbone et al., 2001) and wind turbines (Mangialardi and Mantriota, 1994) by allowing actuators to operate in the most efficient range. Applying this, we investigate and quantify the benefit of using a variable gearhead to improve the efficiency of a robotic leg. The use of a variable gearhead and a Spherical Continuously Variable Transmission (SCVT), as the means to operate an actuator as close to peak efficiency as possible while also delivering the proper output for the required task is introduced. Capehart and Moore have presented the use of CVTs in robotic applications, displaying that these transmissions are viable mechanisms to vary a robots operating characteristics. (Capehart and Moore, 2015). Increasing the task spectrum of a motor by using a CVT as a variable gearhead will also be tested. Simulations of legged locomotion in 1D with a motor model were developed using the force driven-damped SLIP model in MATLAB. The FD-SLIP model was implemented and better results were seen with an optimized control input but motor limitations were run into from the controlled input. The controller used in simulation was the fixed impulse controller. With this, when the body is starting the stance phase the least amount of force was applied to account for maximum compression but for the second half of the phase maximum force is applied to help aid in the amount of thrust into the liftoff event and then touchdown. The touchdown, liftoff, and maximum compression were necessary for the controller. The simulation executed the following equation of motion for the FD-SLIP model configuration: =-g-k0m-l0-blm+um where zeta is the leg length, g is gravity, lo is the force-free leg length, bl is the linear damping term, and m is the mass. This research compares the use of a CVT to the use of a standard gearhead on a traditional robotic actuator in terms of achievable tasks and motor efficiency during operation of these tasks in simulation. The results from this research confirm that the use of a variable gearhead will be beneficial to actuator efficiency during robotic applications and will increase the task range of these actuators.
Funder Acknowledgement(s): National Science Foundation (NSF)
Faculty Advisor: Dr. Carl A. Moore Jr., carl.moore@famu.edu
Role: For this research project, I was responsible for developing and executing 1-D hopping simulations in MATLAB. I was prompted to do the analysis and calculations of the equations of motion of the hopper and to convert them into state space to be used in the ode45 function. I was also responsible for finding the equations that related to the crank-slider mechanism that is on the legged hopper and the jacobian matrix that would help relate the speed of the motor to the torque. I also did some small design modifications to the hopper which included getting rid of old springs and sensors.