An elephant-inspired robotic gripper combines soft actuation with vision-based tactile sensing, enabling robots to grasp fragile objects and perform delicate manipulation with precision.
Researchers at the Japan Advanced Institute of Science and Technology (JAIST) have developed EleTac, a soft robotic gripper inspired by an elephant’s trunk that combines gentle gripping with vision-based tactile sensing and proprioception. Designed to handle fragile and irregularly shaped objects, the lightweight system uses a single internal fisheye camera and artificial intelligence to detect contact, applied force and its own finger position, offering a simpler alternative to sensor-heavy robotic grippers.
Soft robotic grippers have attracted growing interest because they can adapt to objects of different shapes without causing damage. However, integrating tactile sensors into flexible materials has remained a major challenge, as conventional sensors restrict movement or cover only limited areas.
EleTac addresses this by replacing multiple embedded sensors with an internal camera that observes how the gripper deforms during operation. Deep learning algorithms analyse these images to estimate tactile information, allowing the robot to understand contact location, gripping force and finger position simultaneously while maintaining the flexibility of its pneumatic fingers.
During laboratory evaluations, the gripper successfully handled a diverse range of objects, including fruit, tofu, fabric, tools, bolts and playing cards. It also completed demanding tasks such as locating a pen buried in sand using touch alone and cleaning curved tableware by continuously adjusting the applied force.
According to the research team, the compact, lightweight and low-cost design makes EleTac suitable as a plug-and-play end-effector for existing robotic arms and humanoid robots. Beyond industrial automation, the technology could support service robots operating in homes, hospitals and public spaces, where safely handling delicate objects and interacting with people are essential. The research was published in IEEE Transactions on Robotics, highlighting a practical step towards more capable and touch-sensitive soft robotic systems.
