11 Jun Ben-Tzvi lab looks to tails to help robotic movement
The evolution of bipedal robots may follow a road once traveled by living species – the inclusion of a tail.
In the Robotics and Mechatronics Lab of Pinhas Ben-Tzvi, associate professor of mechanical engineering in the College of Engineering, the tail has become a captivating solution for the problem of bipedal and quadrupedal robot stabilizing and maneuvering.
“If you’ve seen robotic quadrupeds, they are very big and very expensive, with articulated legs incorporating multiple degrees of freedom,” said Ben-Tzvi. “The machines use the leg’s multiple degrees of freedom to maneuver and stabilize so if they are pushed from the side, the legs adjust like a human to keep it from falling.”
The problem is that with legs so complex, they are large and very expensive, requiring not only additional joints and motors, but also complex control algorithms and increased computational load. With the addition of a robotic tail, Ben-Tzvi believes the legs can get much simpler, and the robot lighter, easier to design, and less expensive.
“Instead of four leg mechanisms each with multiple degrees of freedom and with complicated controls, we can make the legs with a single degree of freedom, and one motor per leg that will allow the entire mechanism to run forward, but really fast,” explained Ben-Tzvi. “The legs take care of the locomotion, and the tail takes care of stabilization and maneuvering. By decoupling the purposes, we can scale back the complicated legs to make the system much simpler, lighter, more agile, and less expensive.”
The tail works by exerting forces and moments on the robot in six degrees of freedom: forces along the x, y, and z directions and moments about those directions. Ben-Tzvi and his students are mapping the forces and moments generated by the tail motion in an effort to provide stability and maneuvering. The tails are flexible, self-contained and made of the mechanisms that provide the structural backbone of the tail. Actuators and sensors in the tail joints measure position, velocity, and acceleration.
“The idea is that by generating different spatial motions of the tail, we can apply moments and forces around the base of the tail that will result in moments and forces applied to the robot, which will allow it to change direction or maintain stability.”
So far, the team has developed two prototypes using different design principles. The tail prototypes are being used in conjunction with mathematically generated models of legged robots by joining the tails with a computer model of the legged robot and inputting the force of the physical tail into the model to create movement.
“The tail is sitting on a six-axis force sensor that measures the forces and movements and sends that data to the computer where it is applied to a virtual prototype to see how the model affects stabilization and movement,” said Ben-Tzvi. “The modeling is very good at finding the most stable point, allowing us to scale the model and place different masses in different locations along the tail prototype to achieve the desired motion.”
Other researchers have looked at robotic tails that are non-flexible, and only do a single function. Ben-Tzvi said the tail needs to be more robust and flexible, being able to perform a variety of functions to justify incorporating it into a complex, high-performance mechanism.
“We looked to nature to see how animals used their tails and how their tails are structured,” said Ben-Tzvi. “What we saw was a continuous deformation and hyper-redundant structure, so we’ve been inspired by that, and the research we’ve done has been largely establishing the field of hyper-redundant robotic tails. We are pioneering – exploring the field and establishing it so that we have a baseline for coupled dynamic analysis of legged robots with tails onboard.”
After spending a lot of time modeling and establishing the science behind the tails’ potential for usefulness, the team has come up with new mathematical modeling approaches that are widely applicable for flexible structures beyond tails. Because of their multiple mode shapes, the tails allow for a greater range of freedom than single degree of freedom instruments.
“Being able to generate multiple progressive shapes gives us the ability to increase the range of minimum and maximum force and moments that can be generated by the tail,” said Ben-Tzvi. “For a pendulum-like tail the center of mass follows a circular trajectory on a circular arc, but the mode shapes on a flexible tail means the tail can move in a range of positions – in a sphere – allowing it much greater workspace.”
Like evolution, Ben-Tzvi’s tail will take time. With only two hyper-redundant tails, the process to getting them on existing bipedal and quadrupedal robots is not too far down the road. First, Ben-Tzvi will build a quadruped robot of his own to field test the maneuvering and stability properties of the tail, and using data gathered from those experiments, he and his team will discover the best sizes and shapes of tails to fit different robots, and where they need to be placed within the structure to achieve their desired effect.