Abstract
The invention discloses a risk control method and apparatus for demolition robots. During the walking process of a demolition robot, working conditions are determined according to posture information of the demolition robot. A neural network is employed to assess risk categories. When only walking path risks are present, a walking path is re-planned. If only tipping risks are identified, a tipping direction of the demolition robot is determined, and a posture of the demolition robot is adjusted according to the tipping direction. In cases where both walking path risks and tipping risks exist, priority is given to address the tipping risks, followed by the walking path risks. This approach reduces the tipping and walking path risks faced by the demolition robot walking in complex environments, thereby significantly improving the stability and safety of the demolition robot in complex environments.