Robot Path Planning Using Pre-computed Candidate Routes

The patent pre-generates multiple candidate paths using sampling-based path planning before the robot starts operation. These paths are stored and prepared in advance, so when the robot needs to navigate, it can quickly select from pre-computed paths rather than generating new paths in real-time. This preliminary path generation resolves the contradiction by doing the computationally intensive work beforehand, ensuring safety through thorough path planning while enabling fast execution during actual operation.

The patent divides the path planning process into two distinct phases: an offline phase where multiple candidate paths are generated using sampling-based methods, and an online phase where the robot quickly selects from these pre-generated paths. This segmentation separates the time-consuming path generation from the time-critical path selection, allowing the system to maintain high safety standards through comprehensive path planning while achieving real-time responsiveness through efficient path selection from pre-computed options.

The patent performs comprehensive path planning using sampling-based methods that consider robot dynamics and environmental complexity in advance, generating multiple candidate paths that account for various scenarios. This preliminary action ensures that the paths are adapted to complex robot dynamics and environmental conditions before the robot begins operation, resolving the adaptability issue while maintaining fast execution speed during actual navigation.

The patent creates a dynamic path selection system where the robot can switch between pre-generated candidate paths based on real-time environmental conditions and robot state. Instead of using a static grid-based approach, the system maintains multiple pre-computed paths with different characteristics and selects the most appropriate one dynamically, allowing the robot to adapt to changing conditions while benefiting from the speed of pre-computed paths.

The patent pre-generates multiple diverse candidate paths using sampling-based path planning before the robot starts operation. These paths are stored and prepared in advance, providing path diversity for the robot to choose from during actual operation. By doing the diverse path generation in advance, the system achieves both path diversity and fast real-time selection, resolving the contradiction between generating diverse paths and maintaining fast path generation.

The patent creates multiple copies of valid paths with variations through the sampling-based planning process. Instead of regenerating paths each time, the system has multiple pre-computed path copies available for selection. This copying approach provides path diversity while eliminating the need for repeated path generation, thus achieving both diverse path options and fast selection time.