Robotic Vehicle Motion Planning for Static and Dynamic Obstacles

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Autonomous robotic vehicles face challenges in navigating through environments with both static and dynamic objects, as existing methods struggle to accurately predict the movement of dynamic objects and avoid collisions effectively.

Innovation Solution

The implementation of a motion planning module that uses two separate models: a prediction model to assess the movement of dynamic objects and an analytical model to evaluate the presence of static objects, allowing for the combination of scores to determine optimal actions for the robotic vehicle.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a single unified model is used to plan motion for both dynamic and static objects, then the device complexity is reduced, but the measurement precision and reliability of collision avoidance deteriorate because the model cannot accurately predict dynamic object movements while also ensuring safe distances from static objects

Engineering Contradiction:
Improvemodel complexityVSAvoidcollision prediction accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The motion planning module is divided into two separate models: a prediction model that specializes in predicting dynamic object movements using sensor data and machine learning, and an analytical model that specializes in calculating safe distances from static objects using geometric and spatial reasoning. This segmentation allows each model to optimize its algorithms for its specific function, improving overall prediction accuracy while maintaining manageable complexity through modular design.

Inventive Principle:
Principle #1Segmentation

2Adaptability or versatility

If a prediction model using machine learning is used to handle dynamic objects, then the adaptability to unpredictable movements improves, but the reliability and accuracy deteriorate because machine learning predictions are inherently probabilistic and uncertain

Engineering Contradiction:
Improveadaptability to dynamic object movementVSAvoidprediction reliability
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The system merges the probabilistic prediction model with the deterministic analytical model by integrating their outputs in the cost function. The prediction model provides adaptive predictions of dynamic object trajectories, while the analytical model provides reliable geometric constraints. By combining these complementary approaches, the system achieves both adaptability to unpredictable movements and reliability in collision avoidance through the deterministic safety margins.

Inventive Principle:
Principle #5Merging (Combining)

3Reliability

If the robotic vehicle prioritizes avoiding dynamic objects by giving higher weight to prediction model scores, then the safety against collisions with dynamic objects improves, but the productivity and navigation efficiency deteriorate due to excessive caution and reduced speed

Engineering Contradiction:
Improvecollision avoidance reliabilityVSAvoidnavigation efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The cost function applies different weights to different components based on local conditions: the weight for the prediction model component is adjusted based on the presence and proximity of dynamic objects, while the weight for the analytical model component is adjusted based on proximity to static objects. This allows the system to be highly cautious when dynamic objects are present (prioritizing safety) while maintaining efficiency when navigating open spaces (prioritizing productivity), achieving context-dependent optimization.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4148529B1Method for operating a robotic vehicle
Publication Date: 2025.05.14 Y E HUB ARMENIA LLC
  • EP4148529B1 patent drawingFigure 1
  • EP4148529B1 patent drawingFigure 2
  • EP4148529B1 patent drawingFigure 3

AI summary

A method of operating a robotic vehicle is provided. The method includes generating, using a prediction model, first scores, each of the first scores associated with a possible action of the robotic vehicle. The prediction model generates the first scores based, at least in part, on a predicted probability of the robotic vehicle encountering a dynamic object. The method further includes generating, using an analytical model, second scores, each of the second scores associated with a possible action of the robotic vehicle. The analytical model generates the second scores based, at least in part, on the information on the static objects. The method also includes combining the first scores with the second scores to generate combined scores, and selecting an action for the robotic vehicle based, at least in part, on the combined scores. A motion planning module and a robotic vehicle implementing the method are also disclosed.