A robotic system and specifically designed end-effector tool has been developed for the automation of a meat processing task. This tool replicates the operation of a human butcher, while providing improved aseptic conditions and an increase in the reliability, repeatability and quality of the process. The robotic system incorporates a neural network-based fault detection scheme, which uses a learning vector quantisation network to classify axial load signals. A proposed online adaptive learning algorithm will provide continuous optimisation of the robot cut-path, leading to the automation of the path tuning process. This will allow the system to respond to seasonal carcass variations automatically and maintain a consistently high level of performance.