Problem Statement
Mimicking a human-like gait motion in bipedal mechanisms is a problem that has been solved analytically; however, these gait simulations often encounter issues in translation to physical devices. This challenge arises from the number of manufacturing concerns that are difficult to model in simulation such as friction and alignment of the system. These issues, can easily hamper the finely tuned motion preventing movement from being achieved. As a team, we set out to build a bipedal walking mechanism capable of replicating this illusive motion using our knowledge of kinematics and mechanical design.
Project Description
While this problem has been solved and attempted countless time by large companies and researchers alike, success is often achieved using mechatronics controlling a very complex model that achieves balance and consistent motion through programmed weight redistribution and machine learning. Our project took a different approach - we aimed to create a completely mechanical solution to this problem - our goal was to create a machine that would walk using nothing more than mechanisms, a motor, and a power supply.
To create a motion similar to a typical human walking gait, we began with a basic four bar mechanism. We proposed to to create a linkage that would produce an endpoint trajectory similar to the analytically established walking gait. We will discuss the kinematics of this gait in further detail in this writeup. After exploring the capabilities of a four bar linkage, we used Chebyshev's Lamba Mechanism to create the desired D-shaped motion of a bipedal step. Still, there was work to be done creating prototypes that were robust enough to maintain this motion without much departure from Chebyshev's analytical model. Through several iterations, we created a well calibrated Chebyshev mechanism minimizing any play in the joints of the linkage.
Finally, staying true to our goal of solving our problem completely mechanically, we were tasked with developing a counterweight system that would allow us to shift the weight of our robot from side-to-side corresponding with whichever foot was currently on the ground. In this process we developed several iterations to shift a large counterweight from leg to leg. we maintained the following goals: our final design was to be balanced completely through the use of mechanisms, our mechanism was to be powered using one motor, our mechanism would only have one degree of freedom, and finally our mechanism would not be capable of statically balancing without a weight. Many people build bipedal mechanisms that balance using large feet that tesselate as the robot moves forward - these systems balance with ease. we considered this option, but concluded that it was important to our team that we solve and calibrate the balance problem using a mechanical system rather than a change in geometry.
As the result of much iteration, we succeeded in creating a machine that met our team's original goals and purpose. The result of our prototypes and iterative approach will be explained in this write up. We hope to communicate the kinematics of our system, our design process, and our final results for the reader interesting in further exploring the pain-points and possibilities in the development of Bipedal walking mechanisms.