Wall-Climbing Performance of Gecko-inspired Robot with Soft Feet and Digits enhanced by Gravity Compensation

TL;DR

The paper tackles gravity-induced body tilt and unreliable end-effector attachment in inverted-surface climbing by introducing a gecko-inspired, feedforward gravity compensation (FGC) framework. It combines a novel leg-stiffness model with static equilibrium analysis and offline quadratic-programming to coordinate stance legs, computing target foot positions and leg compensations that offset gravity effects. The approach is validated on the EF-I quadruped with soft pneumatic feet, showing a shift from a 3/10 to a 10/10 success rate on inverted surfaces and substantial reductions in roll and pitch fluctuations, while maintaining comparable forward speed. This work enables more reliable adhesion and stable locomotion for small, low-stiffness climbing robots and lays groundwork for real-time, learning-based extensions.

Abstract

Gravitational forces can induce deviations in body posture from desired configurations in multi-legged arboreal robot locomotion with low leg stiffness, affecting the contact angle between the swing leg's end-effector and the climbing surface during the gait cycle. The relationship between desired and actual foot positions is investigated here in a leg-stiffness-enhanced model under external forces, focusing on the challenge of unreliable end-effector attachment on climbing surfaces in such robots. Inspired by the difference in ceiling attachment postures of dead and living geckos, feedforward compensation of the stance phase legs is the key to solving this problem. A feedforward gravity compensation (FGC) strategy, complemented by leg coordination, is proposed to correct gravity-influenced body posture and improve adhesion stability by reducing body inclination. The efficacy of this strategy is validated using a quadrupedal climbing robot, EF-I, as the experimental platform. Experimental validation on an inverted surface (ceiling walking) highlight the benefits of the FGC strategy, demonstrating its role in enhancing stability and ensuring reliable end-effector attachment without external assistance. In the experiment, robots without FGC only completed in 3 out of 10 trials, while robots with FGC achieved a 100\% success rate in the same trials. The speed was substantially greater with FGC, achieved 9.2 mm/s in the trot gait. This underscores the proposed potential of FGC strategy in overcoming the challenges associated with inconsistent end-effector attachment in robots with low leg stiffness, thereby facilitating stable locomotion even at inverted body attitude.

Figures (6)

• Figure 1: (a) The inclination of the quadruped robot’s body with the ambling gait in a 180-degree plane. The orange section highlights the deformation of the links, this leads to the tilting of the body. At this time, when the swing leg attempts to adhere again, not only can the adhesion position not reach the ideal position, but the posture also deviates. (b) The body posture of a dead gecko on the ceiling. (c) The body posture of an alive gecko on the ceiling.
• Figure 2: Overall control framework. The green section represents off-board computation, after the calculation is completed, the offset can be imported into the on-board computation as the preset. The dark blue section represents feedforward compensation for real-time inverse solutions of the gait generator. The gray section represents the parameters input layers, These parameters need to be measured in advance or artificially set according to the motion requirements. The light blue section represents the motors.
• Figure 3: (a) The soft pneumatic foot. When the sole is under negative pressure, the suction cup will generate adhesive force. In addition, the toes will contract to help the adhesion pad attach. Conversely, when the sole is under positive pressure, both the pneumatic and adhesive systems of the sole will lose their ability to adhere. (b) The exploded diagram of the robot's left front leg. The horizontal bevel gear ensures coaxiality through a positioning axis, while the vertical bevel gear is connected to the robot leg, which is connected to a motor outside the body. (c) Schematic diagram of the robot hardware control system. The Micro Controller Unit (MCU) calculates the angle of the servo according to the control flow in Fig. \ref{['fig:diagram']}, and controls the on-state of the positive solenoid valve (PS) and the negative solenoid valve (NS) in the corresponding pneumatic system based on the current motion state of the leg. (d) Quadruped climbing robot EF-I. The coordinate system in the Fig is the body coordinate system.
• Figure 4: (a) A schematic diagram of 180$^\circ$ surface climbing. It should be noted that due to the 180$^\circ$ surface, the robot's left front leg is shown as the right front leg in the picture. $D_m$ represents the x-axis position of the current location relative to the midpoint of the stance diagonal. (b) The diagram shows the z-axis distance that needs to be compensated by the stance legs under different $D_m$ and $\boldsymbol{S}$ in equation \ref{['eq:5']}.
• Figure 5: The temporal diagram for the robot's gait planning. Sampling times are selected to coincide with the phase transition between the swing and stance phases, as indicated by the dashed lines. For the robot's entire motion cycle of duration $T$, the time points are defined as follows: $t_0 = \frac{3}{16}T$, $t_1 = t_0 + \frac{1}{16}T$,$t_i = t_{i-2} + \frac{1}{4}T (i \in \{2, 3, ..., 7\})$. "load" denotes loading force with the attachment action of the footpad through pneumatic mechanisms, while "unload" denotes unloading force with the detachment action of the footpad through pneumatic mechanisms.