Prediction of gripping force of robot-assisted minimally invasive surgery system based on sparrow search algorithm

In response to the insufficient force feedback mechanism in current robot-assisted minimally invasive surgical systems, this study proposes a novel non-sensor indirect clamping force detection method. The relationship between force and compression depth, compression speed, and contact area of the Leaf of a clamp was systematically analyzed through a series of compression experiments conducted on isolated pig stomach tissues. Subsequently, a comprehensive force prediction model is developed by considering multiple influential factors. This model incorporates parameters such as the velocity, displacement, and contact area of the forceps as inputs and employs a backpropagation neural network optimized by Sparrow Search Algorithm (SSA) for precise clamping force prediction, thereby achieving accurate estimation of forces during surgical procedures. The SSA-BP model outperforms both traditional BP models and GA-optimized BP models in terms of prediction accuracy and other performance indicators, showcasing its superior performance. After optimization and improvement through SSA, the determination coefficient between this model and experimental clamping force reaches 0.9954, representing an increase of 0.83% and 1.41% compared to the GA-optimized BP model and traditional BP neural network model, respectively. The SSA-BP model proposed in this study demonstrates superior predictive ability and a higher degree of fitting, thereby offering a dependable clamping force control scheme for robot-assisted minimally invasive surgery systems.