Pedestrian Head-Pose Detection for Occluded Vehicle Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating due to the vast amount of data they need to process and store, including image data, map data, and sensor data, which can limit their navigation capabilities and increase the complexity of decision-making.
Innovation Solution
The use of cameras to analyze images and generate indicators of occluded pedestrians, allowing the vehicle to make navigational decisions based on the contact position of the pedestrian with the ground surface, combined with a sparse map for navigation that reduces data storage and transfer requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional mapping technology is used for autonomous vehicle navigation, then the vehicle can navigate using existing map data, but the sheer volume of data needed to store and update the map poses daunting challenges
Solution Approach 1:
The patent extracts only the essential navigational elements from traditional maps, creating a sparse map that contains only critical information needed for navigation. This reduces data storage requirements while maintaining navigation reliability by focusing on key features rather than storing complete traditional map data.
Solution Approach 2:
The navigation system segments the environment into relevant and irrelevant components, processing only the necessary visual information for navigation decisions. This segmentation allows the vehicle to navigate effectively without processing or storing the entire visual environment, reducing data volume while maintaining navigation capability.
2Measurement precision
If vast volumes of information are collected and analyzed by the autonomous vehicle, then the vehicle can make informed navigation decisions, but the sheer quantity of data poses challenges that can limit or adversely affect autonomous navigation
Solution Approach 1:
The system applies local quality by focusing computational resources on analyzing only those portions of the visual environment that are relevant to navigation decisions. Rather than uniformly processing all captured data, the system identifies and analyzes specific regions of interest, reducing processing complexity while maintaining perception accuracy for critical elements.
Solution Approach 2:
The patent implements partial action by processing only the necessary subset of visual data required for safe navigation. Instead of analyzing every pixel and object in the environment, the system selectively processes information that directly impacts navigation decisions, reducing computational complexity while maintaining sufficient environmental awareness.
3Reliability
If the autonomous vehicle uses cameras to capture images of occluded pedestrians, then the vehicle can identify pedestrians partially hidden from view, but the captured images lack complete information about the pedestrian's contact position with the ground surface
Solution Approach 1:
The system performs preliminary analysis of captured images to identify occluded pedestrians and predict their contact positions before making navigation decisions. By pre-processing the visual data to estimate contact positions of partially visible pedestrians, the system recovers lost information and improves detection reliability without requiring complete visual information.
Solution Approach 2:
The patent introduces an intermediary computational process that infers missing contact position information from partial visual data. This intermediary analysis layer bridges the gap between incomplete image data and the needed contact position information, allowing the system to detect occluded pedestrians accurately despite information loss in the captured images.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Systems and methods are provided for vehicle navigation. In one implementation, a navigation system for a host vehicle may include at least one processor programmed to receive from a camera onboard the host vehicle at least one captured image representative of an environment of the vehicle; detect a pedestrian represented in the at least one captured image; analyze the at least one captured image to determine an indicator of angular rotation and an indicator of pitch angle associated with a head of the pedestrian represented in the at least one captured image; and cause at least one navigational action by the host vehicle based on the indicator of angular rotation and the indicator of pitch angle associated with the head of a pedestrian.