Open Door Prediction Neural Network for Autonomous Vehicle Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles struggle to accurately predict whether nearby vehicles have open doors, which are indicative of potential human presence or stationary status, impacting safe navigation and decision-making.
Innovation Solution
An open door prediction neural network that processes sensor data, including point clouds from laser sensors and images from cameras, to generate likelihood scores for open doors, integrating segmentation tasks to enhance classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human-programmed logic is used to make full-vehicle predictions, then the system can precisely specify how sensor outputs are combined and transformed, but the system lacks the ability to accurately predict open door states and human presence
Solution Approach 1:
The patent replaces human-programmed logic with a neural network model that automatically learns prediction rules from training data. The neural network processes sensor data (point clouds, images) to predict open door states and human presence, substituting manual programming with adaptive machine learning that improves detection accuracy without requiring explicit rule specification.
2Reliability
If the autonomous vehicle planning system considers open door predictions, then safety is improved by avoiding hazards, but the system complexity increases due to integration of additional prediction tasks
Solution Approach 1:
The patent merges the open door prediction task with the full-vehicle prediction task in a unified neural network model. The network simultaneously predicts vehicle state, open door status, and human presence from the same sensor inputs, consolidating multiple functions into a single system that improves safety without proportionally increasing complexity.
Solution Approach 2:
The neural network model is designed to perform multiple functions: full-vehicle prediction, open door detection, and human presence inference. This multi-functional approach allows the system to derive safety-critical information from a single prediction framework, reducing the need for separate specialized systems.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for open vehicle doors prediction using a neural network model. One of the methods includes: obtaining sensor data (i) that includes a portion of a point cloud generated by a laser sensor of an autonomous vehicle and (ii) that characterizes a vehicle that is in a vicinity of the autonomous vehicle in an environment; and processing the sensor data using an open door prediction neural network to generate an open door prediction that predicts a likelihood score that the vehicle has an open door.