Pedestrian Motion Prediction Using Map and Social Context

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for predicting pedestrian velocity vectors and poses in self-driving vehicles do not adequately consider future changes in pedestrian movement due to interactions with stationary obstacles and social dynamics, leading to potential collisions.

Innovation Solution

A method that incorporates a static map of the prediction area, current velocity vectors of other pedestrians, and social interactions to predict future pedestrian movements, grouping pedestrians and accounting for their destinations and social behaviors to enhance prediction accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing methods only use current pedestrian position and velocity vector for prediction, then the prediction process is simple and fast, but the prediction accuracy is insufficient leading to potential collisions

Engineering Contradiction:
Improveprediction accuracyVSAvoidprediction process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The prediction process is segmented into multiple independent modules: a map module that processes static environmental data, a social context module that analyzes pedestrian interactions, and a prediction module that integrates these factors. This segmentation allows each module to specialize in specific aspects of prediction while maintaining overall system manageability and accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs preliminary processing of map data and social context information before the actual prediction occurs. The map module pre-processes static environmental features, and the social context module pre-identifies interaction patterns, so that when prediction is needed, these pre-computed data are readily available to enhance accuracy without adding computational burden during critical prediction moments.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If the prediction considers social interactions and environmental factors, then collision avoidance capability is improved, but the computational time and processing load increase

Engineering Contradiction:
Improvecollision avoidance capabilityVSAvoidcomputational time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system applies different levels of analysis to different aspects of pedestrian behavior. Social interactions and environmental factors are considered with appropriate detail only where relevant to the specific prediction scenario, rather than uniformly applying complex analysis to all cases. This local quality approach maintains high reliability for critical predictions while reducing unnecessary computational overhead.

Inventive Principle:
Principle #3Local quality

3Measurement precision

If pedestrians are tracked individually with full social context analysis, then prediction accuracy is improved, but the system complexity and data processing requirements increase significantly

Engineering Contradiction:
Improvepedestrian movement prediction accuracyVSAvoiddata processing requirements
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The map module and social context module are merged into an integrated prediction framework where static environmental data and dynamic social interaction data are combined and processed together. This merging allows the system to leverage correlations between environmental features and social behaviors, improving prediction accuracy while avoiding the redundancy of separate independent analysis systems.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP3844664B1Method for predicting at least one future velocity vector and/or a future pose of a pedestrian
Publication Date: 2025.12.10 ROBERT BOSCH GMBH
  • EP3844664B1 patent drawingFigure 1
  • EP3844664B1 patent drawingFigure 2
  • EP3844664B1 patent drawingFigure 3~4

AI summary

The invention relates to a method for predicting (30) at least one future velocity vector and/or a future pose of a pedestrian in an area of prediction. A map of the surroundings of the pedestrian and current velocity vectors of other pedestrians in the area of prediction are taken into account for the prediction (30).