Autonomous Vehicle Motion Planning Using Probabilistic Control Input Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle control systems face challenges in computing safe and efficient vehicle motions due to noisy sensor data and uncertainties in vehicle and environmental parameters, leading to computational complexity and potential nondrivable paths.
Innovation Solution
The method evaluates the quality of control inputs by determining the probability of moving the vehicle into a desired state, sampling control inputs, and mapping them to control commands, rather than evaluating the entire motion from initial to target state, using a probabilistic model to account for uncertainties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire motion from initial state to target state is evaluated to determine quality, then the quality of motion is determined, but the computational load becomes overwhelming and the tree of possible motions becomes unnecessarily large
Solution Approach 1:
The patent segments the motion evaluation process into two parts: evaluating only the control input quality (transition from one state to another) rather than evaluating the entire motion trajectory. This segmentation reduces the computational complexity from O(n) where n is the number of states in the trajectory to O(1) for each control input evaluation, making real-time computation feasible while maintaining reliability.
Solution Approach 2:
The patent extracts and focuses only on the critical element for quality determination - the control input that transitions the vehicle from one state to another. By taking out the unnecessary evaluation of intermediate states and focusing solely on the control input quality, the system achieves both computational efficiency and motion quality assurance.
2Productivity
If random configurations or control inputs are generated and connected to grow a tree of possible motions, then computational complexity is reduced, but the path may be nondrivable
Solution Approach 1:
The patent introduces feedback by evaluating the quality of each control input using a cost function that checks whether the control input produces a valid state transition. This feedback mechanism ensures that only drivable paths are selected, as control inputs that would lead to nondrivable paths are identified and rejected through the quality evaluation process.
Solution Approach 2:
The patent applies dynamics by using a dynamic model of the vehicle to predict the next state based on the current state and control input. This dynamic evaluation ensures that the generated paths are physically realizable and drivable, as the model accounts for vehicle constraints and dynamics rather than treating all state transitions as equally valid.
3Reliability
If the tree of possible motions is made large to ensure quality motion, then the quality of motions is improved, but the computational and memory load becomes overwhelming
Solution Approach 1:
The patent segments the motion planning problem into evaluating individual control inputs rather than storing and evaluating entire motion trajectories. This segmentation allows the system to maintain high motion quality standards while using minimal memory, as only the current control input and its immediate effect need to be evaluated, not the entire tree of possible motions.
Solution Approach 2:
The patent changes the evaluation parameter from assessing entire motion trajectories to assessing control input quality. This parameter change fundamentally reduces the memory requirements, as the system only needs to store and evaluate control inputs rather than complete motion paths, achieving both high quality and low memory consumption.
Data Source
AI summary
A method controls a motion of a vehicle using a model of the motion of the vehicle that includes an uncertainty. The method samples a control space of possible control inputs to the model of the motion of the vehicle to produce a set of sampled control inputs and determines a probability of each sampled control input to move the vehicle into state satisfying constraints on the motion of the vehicle. The method determines, using the probabilities of the sampled control inputs, a control input having the probability to move the vehicle in the state above a threshold. The control input is mapped to a control command to at least one actuator of the vehicle to control the motion of the vehicle.