<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
Inspired by the dynamic coupling of moto-neurons and physical elasticity in animals, this work explores the possibility of generating locomotion gaits by utilizing physical oscillations in a soft snake by means of a low-level spiking neural mechanism. To achieve this goal, we introduce the Double Threshold Spiking neuron model with adjustable thresholds to generate varied output patterns. This neuron model can excite the natural dynamics of soft robotic snakes, and it enables distinct movements, such as turning or moving forward, by simply altering the neural thresholds. Finally, we demonstrate that our approach, termed SpikingSoft, naturally pairs and integrates with reinforcement learning. The high-level agent only needs to adjust the two thresholds to generate complex movement patterns, thus strongly simplifying the learning of reactive locomotion. Simulation results demonstrate that the proposed architecture significantly enhances the performance of the soft snake robot, enabling it to achieve target objectives with a 21.6% increase in success rate, a 29% reduction in time to reach the target, and smoother movements compared to the vanilla reinforcement learning controllers or Central Pattern Generator controller acting in torque space.
...
Inspired by the dynamic coupling of moto-neurons and physical elasticity in animals, this work explores the possibility of generating locomotion gaits by utilizing physical oscillations in a soft snake by means of a low-level spiking neural mechanism. To achieve this goal, we introduce the Double Threshold Spiking neuron model with adjustable thresholds to generate varied output patterns. This neuron model can excite the natural dynamics of soft robotic snakes, and it enables distinct movements, such as turning or moving forward, by simply altering the neural thresholds. Finally, we demonstrate that our approach, termed SpikingSoft, naturally pairs and integrates with reinforcement learning. The high-level agent only needs to adjust the two thresholds to generate complex movement patterns, thus strongly simplifying the learning of reactive locomotion. Simulation results demonstrate that the proposed architecture significantly enhances the performance of the soft snake robot, enabling it to achieve target objectives with a 21.6% increase in success rate, a 29% reduction in time to reach the target, and smoother movements compared to the vanilla reinforcement learning controllers or Central Pattern Generator controller acting in torque space.
Living beings are extremely adept at executing complex and dexterous manipulation skills by integrating tactile, visual, and other stimuli. Robotics researchers aims to endow the robots with similar manipulation intelligence. From robotics and machine learning domains, although recent years we have seen lots of promising results on visual imitation/exploration learning for robot manipulation. e.g., the robot can learn the adaptive behavior from the trajectories of human demonstrations. However, these approaches face challenges in generalizing to diverse tasks, especially for the tasks involving contact. To this end, bunch of approaches have been developed exploiting the contact and adaptive force control in a compensation way. While these ad-hoc solutions are practical for implementing specific functionalities, they fall short of providing a comprehensive scientific understanding of manipulation, we have to figure out a unified framework to unveil the mystery of the manipulation. Living beings can systemically combine these two works together and finish the given task in a smooth, safe and intelligent way. This makes us believe that they have a special capability/mechanism to learn, generalize and control the complex manipulation exploiting their multi-modality feedback which we call dexterity intelligence. Understanding and evaluating the dexterity intelligence are not trivial, it needs input from different research domains. In this special issue, we accept 8 papers and hope that they can partially unveiling the mystery of “dexterity intelligence”. [...]
...
Living beings are extremely adept at executing complex and dexterous manipulation skills by integrating tactile, visual, and other stimuli. Robotics researchers aims to endow the robots with similar manipulation intelligence. From robotics and machine learning domains, although recent years we have seen lots of promising results on visual imitation/exploration learning for robot manipulation. e.g., the robot can learn the adaptive behavior from the trajectories of human demonstrations. However, these approaches face challenges in generalizing to diverse tasks, especially for the tasks involving contact. To this end, bunch of approaches have been developed exploiting the contact and adaptive force control in a compensation way. While these ad-hoc solutions are practical for implementing specific functionalities, they fall short of providing a comprehensive scientific understanding of manipulation, we have to figure out a unified framework to unveil the mystery of the manipulation. Living beings can systemically combine these two works together and finish the given task in a smooth, safe and intelligent way. This makes us believe that they have a special capability/mechanism to learn, generalize and control the complex manipulation exploiting their multi-modality feedback which we call dexterity intelligence. Understanding and evaluating the dexterity intelligence are not trivial, it needs input from different research domains. In this special issue, we accept 8 papers and hope that they can partially unveiling the mystery of “dexterity intelligence”. [...]
