Automated Driving Path Planning Using Probabilistic Object Likelihood
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated driving vehicles face challenges in determining a path when there are many moving objects, as predicted object positions are distributed over a wide range, making it difficult to establish a viable path.
Innovation Solution
A vehicle control apparatus that acquires information on the surrounding environment, calculates probabilities of object presence using a first distribution, and combines this with travel data from a predetermined driver's model to determine a suitable path, even in regions without defined models by sandwiching the area between two defined regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predicted positions of moving objects are calculated to ensure safety, then the distribution range of predicted positions becomes wide, but this makes it difficult to determine a viable path for the vehicle
Solution Approach 1:
The patent changes the parameter of path determination from binary (safe/unsafe) to probabilistic (likelihood values). By calculating the likelihood that a position is safe based on the probability distribution of object positions, the system can identify viable paths even when object positions are uncertain. This transforms the wide distribution of predicted object positions from a problem into useful information for path planning.
Solution Approach 2:
The patent introduces an intermediary element: the likelihood calculation unit that mediates between the predicted object positions and the path determination. This intermediary computes the probability that each position is safe by considering the overlap between the vehicle's predicted position and the object's predicted position distribution, thereby resolving the contradiction between safety and path feasibility.
2Measurement precision
If multiple models are used to cover all regions, then path determination accuracy improves, but the system complexity and computational burden increase
Solution Approach 1:
The patent makes the existing models universal by enabling them to evaluate any region through the likelihood calculation mechanism. Instead of requiring separate models for each region, the system uses the probability distribution framework to evaluate safety likelihoods across all regions uniformly, reducing system complexity while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-calculating the probability distributions of object positions and storing them. When path determination is needed, the system directly uses these pre-computed distributions to calculate likelihoods, avoiding the need for complex real-time calculations and reducing computational burden while maintaining accuracy.
Data Source
AI summary
A vehicle control apparatus for controlling automated driving of a vehicle acquires information relating to a situation in a surrounding area of the vehicle, acquires, for each position, a first value relating to a probability that an object present in the surrounding area will be present at a future point in time, and a second value obtained based on travel data of a predetermined driver based on the information, the second values being specified using a model defined for a portion of regions through which the vehicle travels, and if the vehicle travels through a region for which a model is not defined, the second values being each specified by combining two values acquired using two models defined for two regions sandwiching the position through which the vehicle is traveling, and determines a path on which the vehicle is to move.