Pedestrian Probability Ellipse Prediction for Autonomous Vehicles

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Autonomous driving vehicles face challenges in accurately predicting pedestrian movements, especially in low-speed scenarios like parking lots and campus roads, where interactions with pedestrians are frequent but prediction accuracy is inconsistent.

Innovation Solution

A computer-implemented method for autonomous driving vehicles that receives images from onboard cameras, identifies obstacles, predicts their location at multiple time points, and generates a probability ellipse with varying probability indicators within the ellipse, aiding in trajectory planning to navigate around pedestrians safely.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional prediction methods are used for pedestrian movement, then the system complexity remains low, but the prediction accuracy deteriorates in low-speed scenarios with frequent pedestrian interactions

Engineering Contradiction:

Improvepedestrian movement prediction accuracyVSAvoidprediction system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The prediction system is segmented into multiple independent modules: probability ellipse generation module, trajectory prediction module, and interaction strategy module. Each module handles specific aspects of pedestrian prediction, allowing the system to achieve high accuracy through specialized sub-systems while maintaining manageable overall complexity through modular architecture.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system transitions from traditional point-based or box-based prediction to ellipse-based probability distribution prediction. This dimensional change allows the system to represent not just predicted locations but also uncertainty and probability distributions, significantly improving prediction accuracy by capturing the probabilistic nature of pedestrian movements in complex environments.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Measurement precision

If probability ellipse prediction is implemented, then the prediction accuracy improves, but the computational complexity increases

Engineering Contradiction:

Improvepedestrian location prediction accuracyVSAvoidcomputational power requirement

Core Design Contradiction:

Measurement precisionVSPower

Solution Approach 1:

Instead of computing complex probability distributions across the entire scene, the system applies probability ellipse prediction locally to each detected pedestrian. Each pedestrian receives an individual probability ellipse based on their specific movement patterns and context, reducing overall computational load while maintaining high local prediction accuracy for each target.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system pre-computes probability ellipse parameters and trajectory predictions for multiple future time points before actual navigation decisions are required. This preliminary action allows the autonomous vehicle to have ready-made prediction data for rapid decision-making, reducing real-time computational power requirements while maintaining high prediction accuracy.

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If interaction strategies for urban roads are applied to low-speed scenarios, then the operational framework remains simple, but the prediction accuracy deteriorates due to different pedestrian behaviors

Engineering Contradiction:

Improveprediction system adaptability to different scenariosVSAvoidpedestrian movement prediction accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system dynamically adapts its prediction model based on the operational scenario. For low-speed scenarios with frequent pedestrian interactions, the system activates enhanced interaction strategies that incorporate scenario-specific parameters and behavioral patterns. This dynamic adaptation allows the system to maintain high prediction accuracy across different environments without requiring completely separate operational frameworks.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes key prediction parameters based on scenario type. In low-speed scenarios, parameters such as prediction time horizon, probability distribution characteristics, and interaction weightings are adjusted to reflect the different nature of pedestrian behaviors. This parameter adaptation enables accurate prediction across diverse scenarios while maintaining a unified underlying prediction framework.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11230297B2Pedestrian probability prediction system for autonomous vehicles

Publication Date: 2022.01.25 BAIDU USA LLC

AI summary

According to some embodiments, a system receives a captured image perceiving an environment of an ADV from an image capturing device of the ADV. The system identifies an obstacle in motion near the ADV based on the captured image. The system predicts a location for the moving obstacle at each of a number of time points. The system generates a probability ellipse based on the predicted location at the each time point, where the probability ellipse includes a probability indicator indicating different probabilities of the moving obstacle for different locations within the probability ellipse at the each time point.