A new robotic system named Sashimi-Bot uses AI and tactile sensing to autonomously prepare fish fillets, demonstrating advanced manipulation of deformable objects.
Researchers have developed a novel three-armed robotic system, dubbed Sashimi-Bot, capable of autonomously preparing fish fillets. This development addresses the significant challenges robots face when handling slippery, deformable objects like fish, which can change shape during manipulation and cutting.
The Sashimi-Bot system features specialized arms for distinct tasks: one arm gently straightens the fish, a second arm wields a knife for slicing, and a third arm uses chopsticks to transfer the slices to a tray. This tri-manual approach mimics human dexterity in food preparation.
Instead of relying on traditional programming, the system employs artificial intelligence. The robot underwent extensive training in a virtual simulation, learning through trial and error how to optimally handle and straighten the fish. For the cutting arm, machine learning was utilized to interpret data from a touch-sensitive sensor near the knife, providing a sense of touch.
Further enhancing its precision, a separate computer model was trained to detect contact with the cutting board. This allows the robot to adjust its cutting trajectory in real time, preventing excessive pressure or damage to the cutting surface. When tested with real salmon loins, the Sashimi-Bot successfully performed all tasks, albeit at a slower pace than a human chef, demonstrating its capability in manipulating and slicing delicate, slippery materials.
This development is significant for robotic manipulation of soft, deformable objects, a common challenge in food processing and beyond. The integration of AI for learning complex tasks and tactile sensing for real-time control represents a leap in robotic dexterity. These advancements could be transferable to other industries requiring precise handling of delicate materials, such as in medical applications or advanced manufacturing processes.
Edited by the news editor with AI from the original report — please refer to the original source.