Impact-resistant, autonomous robots inspired by tensegrity architecture

Abstract

Future robots will navigate perilous, remote environments with resilience and autonomy. Researchers have proposed building robots with compliant bodies to enhance robustness, but this approach often sacrifices the autonomous capabilities expected of rigid robots. Inspired by tensegrity architecture, we introduce a tensegrity robot -- a hybrid robot made from rigid struts and elastic tendons -- that demonstrates the advantages of compliance and the autonomy necessary for task performance. This robot boasts impact resistance and autonomy in a field environment and additional advances in the state of the art, including surviving harsh impacts from drops (at least 5.7 m), accurately reconstructing its shape and orientation using on-board sensors, achieving high locomotion speeds (18 bar lengths per minute), and climbing the steepest incline of any tensegrity robot (28 degrees). We characterize the robot's locomotion on unstructured terrain, showcase its autonomous capabilities in navigation tasks, and demonstrate its robustness by rolling it off a cliff.

Figures (15)

• Figure 1: An impact-resistant tensegrity robot exhibits autonomous locomotion in an unstructured environment. A 3-bar tensegrity robot rolls off a 2 m cliff and survives the crash landing due to its inherent body compliance. Afterward, it continues its locomotion in the same direction. The robot uses on-board sensors to estimate its shape and orientation. Estimating its cable lengths, downward face, and heading allows the robot to continue its locomotion after the disturbance caused by the initial fall.
• Figure 2: Real-time state estimation. (A) A locomotion trial in a field setting with photographs and corresponding renderings of the robot's estimated state from onboard sensors. (B) A plot of the same locomotion trial showing the results of state estimation (SE) from onboard sensors compared to ground truth (GT) from our vision-based pose tracking algorithm lu20226n. The top subplot shows the nine tendon lengths, the middle subplot shows the ZYX Euler angles that describe the robot's orientation relative to the global frame defined by Earth's gravitational and magnetic fields, and the bottom subplot shows the root mean square error (RMSE) between the estimated and ground truth positions of the nodes relative to the bar length. The robot executes its counterclockwise turning gait.
• Figure 3: Locomotion Summary. The tensegrity robot can achieve locomotion across unstructured terrains, including (A) grass, (B) ice, (C) pebbles, and (D) sand. It can also climb (E) inclines as steep as 28°. Its top speed is achieved with its (F) dynamic rolling gait. It can also turn (G) counterclockwise and (H) clockwise. Each locomotion experiment is shown as a stitched photograph and a corresponding plot. Arrows indicate the locomotion direction. Each plot shows four trials of three cycles of the corresponding gait. The locomotion speed is compared (I) across terrains, (J) as a function of incline angle (see Fig. S6), (K) for different rolling gaits, and (L) for different turning gaits (see Fig. S9). The bar plots show the average speed over four trials while error bars represent the maximum and minimum speeds.
• Figure 4: Autonomous control. (A)-(C) The tensegrity robot autonomously follows a provided trajectory via a model-based controller. The robot plans a sequence of actions---shown with black markers for the predicted center of mass (CoM) and magenta vectors for the predicted principal axis---that enable it to best follow the trajectory plotted in cyan. The darkest magenta vector corresponds to the robot's measured starting state and each lighter vector to the successive states predicted for following the planned action sequence. Each corresponding plot shows the target trajectory and robot's CoM for three trials. Circular markers indicate the CoM at planning steps. The tracked trajectories include (A) a straight line, (B) a circular arc, and (C) a right triangle. (D) The tensegrity robot autonomously plays limbo, sensing the height of the yellow bar with an off-board camera and adapting its body height to roll under it. The corresponding plot shows the height of the robot compared to the bar height during the limbo demonstration. The gray shaded regions show when the robot is executing its rolling gait and when it is waiting for the bar to lower while the yellow shaded regions show when part of the robot is directly under the bar.
• Figure 5: Impact resistance. The robot survives a 5.7 m drop after rolling off a bridge. Movie S8 shows the robot continuing its locomotion after surviving the fall.