Driving Road Estimation Using Probabilistic Road Occupancy Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic vehicle following systems face challenges in maintaining accurate tracking of target moving objects during severe traffic conditions and night scenarios due to frequent changes in light conditions and the presence of other vehicles or pedestrians, leading to unstable tracking results and increased risk of collision.
Innovation Solution
A method and system for predicting a road occupancy map of a target moving object by discretizing target status information into a Gaussian distribution, sampling discrete values, and calculating probability values to estimate the road region the vehicle will travel in, allowing safe navigation without relying on the surrounding environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If vehicle-mounted radar and camera are used for tracking, then automatic vehicle following is achieved, but tracking stability deteriorates in severe traffic conditions and night scenarios
Solution Approach 1:
The patent introduces a road occupancy map as an intermediary representation that decouples the tracking system from direct dependence on radar and camera detections. Instead of relying solely on these sensors to track individual vehicles, the system uses the occupancy map to predict safe road regions, serving as a mediator that stabilizes tracking in severe conditions where direct sensor tracking fails.
Solution Approach 2:
The patent replaces the mechanical/optical tracking system (radar and camera-based detection) with a predictive modeling approach using road occupancy maps. This substitution transitions from active sensing and tracking to predictive path planning based on probability distributions, eliminating the instability inherent in direct sensor tracking during night scenarios and severe traffic conditions.
2Loss of information
If tracking relies on surrounding environment detection, then target identification is possible, but accuracy deteriorates when traffic conditions are indefinite
Solution Approach 1:
The patent performs preliminary action by predicting future road occupancy and safe travel regions before actual tracking failures occur. By using the road occupancy map to anticipate safe regions ahead, the system proactively maintains tracking accuracy without waiting for environmental conditions to deteriorate, thus preventing information loss before it happens.
Solution Approach 2:
The patent changes the parameter representation from direct target coordinates (which become unreliable in indefinite conditions) to probability distributions of road occupancy. This parameter transformation allows the system to maintain measurement precision by working with statistical likelihoods rather than deterministic positions that fail under severe traffic and night conditions.
3Productivity
If discrete values are sampled from Gaussian distribution, then calculation efficiency is improved, but estimation precision may be reduced
Solution Approach 1:
The patent applies partial action by sampling only the necessary discrete values from the Gaussian distribution that are sufficient for road region estimation, rather than computing the entire continuous distribution. This selective sampling achieves adequate estimation precision for safety-critical applications while dramatically improving calculation efficiency for real-time autonomous driving decisions.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Embodiments of the present invention disclose a driving road estimation method and a driving road estimation system. The method in the embodiments of the present invention includes the following steps. When obtaining target status information of a surrounding target moving object, a vehicle may perform discretization processing on the target status information to obtain a parameter sequence, so that the vehicle can perform prediction based on the target status information and a target discrete value that is included in the parameter sequence, to obtain a target movement track. The vehicle can determine, based on the target movement track, a road region in which the target moving object travels at a subsequent moment, and then the vehicle can determine a region in which the vehicle travels at the subsequent moment, so that the region in which the vehicle travels at the subsequent moment does not overlap the region in which the target moving object travels at the subsequent moment. According to the method described in the embodiments, the vehicle safely travels in the estimated road region, thereby effectively improving driving safety and reliability of the vehicle.