biomimetic robotic tails for stabilising and maneuvering legged robots

This research aims to study the effectiveness of utilizing a robotic tail to stabilize and maneuver legged robots. Shortcomings of current legged robot locomotion include a limited means of adjusting the robot's center of mass, and limited dynamic responsiveness at the leg-ground contact. Utilizing a continuum robotic tail may have the potential to solve both of these issues: the positioning of the tail could be used to adjust the robot's center of mass, and the robot's dynamic responsiveness would be improved by the tail's ability to quickly adjust shape.

 

In addition, the inclusion of a continuum tail would enable different paradigms of locomotion. For example, in biped robots, the primary focus has been human-like robots where the torso is held vertical during movement, as shown in Fig.1(a). The tail will enable an alternative dinosaur-inspired approach, shown in Fig. 1(b), where the robot's torso is biased forward and the tail balances it behind the legs.

 

Typical scenarios for which a robotic tail could be used in place of modifying the robot's gait include stabilizing the robot after it walks over a ledge (Fig. 2(a)), turning the robot (Fig. 2(b)), or stabilizing the robot in response to a lateral ground shift (Fig. 2(c)).

 

Several design paradigms are being investigated as candidates to accomplish these aims. One example includes a novel hybrid continuum robot design (shown in Fig. 3) utilizing both rods and cables to transmit actuation. It is believed that this approach may balance the benefits of cable-driven continuum structures, such as lower response times, and rod-driven structures, such as increased dynamic stability, in a single system.

 

There are four key aspects of this research: tail mechanics, sensing, task planning and control, shown together in Fig. 4. The method of virtual power is being used to create static and dynamic mechanics models of the continuum tail, and will include features such as models for the actuation mechanism and dynamic friction. These models will assist in the tail's design and optimization, as well as simulation-based testing of the task planning algorithms. Integrated sensing will enable the continuum tail to estimate its shape in real-time from multiple sources of sensor data, including actuation force measurement, rod/cable displacements and tail accelerometers and gyroscopes. Task planning will enable generation of tail trajectories that will provide the required loading (force and/or moment) to stabilize or maneuver the mobile robot.  Control will implement these desired tail trajectories, utilizing the real-time shape sensing to ensure the desired tail trajectory is followed.

 

The long-term goal of this project is to fully integrate a robotic tail with biped and quadruped legged robots and demonstrate their effectiveness at stabilizing and maneuvering the physical system. In the short-term, a continuum tail experimental test platform will be constructed and mounted on a six-axis load cell. The real-time measurements from this load cell will be coupled to a dynamic simulation of the legged robot in MSC ADAMS. This 'hardware-in-the-loop' testing will enable demonstration of the tail's effectiveness in a more controlled environment for the early stages of this research.

 


Figure 1. Bipedal locomotion of: (a) a humanoid, with the torso held above the body, (b) a Tyrannosaurus, with the torso extended in front of the legs, and the tail extended behind.

 

 


Figure 2. Illustrations of case studies for impulsive actions: (a) traversing a vertical ledge; (b) sharp turning with small radii while moving, (c) stabilizing after a sudden ground displacement while moving. Black arrows represent sample motion path, and red arrows represent applied forces. Inset axes in each figure illustrate the moment direction to be considered: pitch (a), yaw (b), and roll (c).

 

 


Figure 3. Illustration of the tail experimental platform: a flexible tail and its actuation module are rigidly mounted on a six-axis load cell, with a vision system to measure the shape of the continuum tail using visual markers, with on-board sensing embedded along the tail and within the actuation module.

 


Figure 4. Key aspects of this research: tail mechanics, sensing, task planning and control

 

 

 

 

 

 

Copyright © 2018 Robotics and Mechatronics Laboratory (RML), 635 Prices Fork Road, Blacksburg, VA 24061