Cutting Force in Pork Carcass Operation Robots

What is it about?

In the process of dividing pork carcasses by robots, the resistance of knives varies randomly due to the uneven distribution of meat density. The conventional pork carcass robot is unable to adjust the cutting force and cutting angle according to the meat density distribution during meat segmentation, which leads to low meat segmentation quality and damage to the tool.

Featured Image

Photo by Frank Zhang on Unsplash

Photo by Frank Zhang on Unsplash

Why is it important?

To address the above problems, this paper proposes a reinforcement meta-learning based cutting force with shape regulation method. First, a reinforcement learning-based cutting tool shape-following regulation model is constructed, and the segmentation task sequence is reinforced and trained to obtain the optimal action sequence to improve the robot's adaptability for different tasks. Secondly, the stochastic gradient descent method is used to train the reinforcement learning action sequence parameter changes to improve the generalization ability of the algorithm. Finally, the combination of meta-learning and LSTM achieves long-term memory for long action sequences, and at the same time can effectively solve the gradient explosion and gradient disappearance problems in the training process.

Perspectives