Micro-Mobility Risk Profiling From Rider Appearance and Movement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI-based models are ill-equipped to determine the risk posed by users of micro-mobility vehicles, such as bicycles and scooters, when predicting their behavior, as they do not consider attributes of these vehicles, leading to inaccurate predictions in autonomous and semi-autonomous vehicle systems.
Innovation Solution
A machine learning-based system that uses trained models to extract and classify micro-mobility vehicles and their riders from video streams, generating comprehensive risk profiles that include characteristics like vehicle type, movement patterns, and user appearance to predict potential actions and risks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current AI-based models are used to predict micro-mobility rider behavior, then the system can process video streams and generate predictions, but the predictions are inaccurate because the models do not consider attributes of micro-mobility vehicles
Solution Approach 1:
The system segments the rider detection task into multiple independent detection models: one for detecting the rider, another for detecting the micro-mobility vehicle, and a third for determining the relationship between them. This segmentation allows each model to specialize in specific attributes, improving overall prediction accuracy while maintaining system adaptability.
Solution Approach 2:
The system adds a new dimension of analysis by explicitly detecting and incorporating micro-mobility vehicle attributes (type, configuration, capabilities) as separate features in the risk profile. This dimensional expansion transforms the prediction model from considering only rider characteristics to considering the combined rider-vehicle system, thereby improving prediction accuracy.
2Measurement precision
If comprehensive risk profiles including vehicle attributes are generated, then prediction accuracy improves, but system complexity increases
Solution Approach 1:
The system divides the complex risk assessment task into separate modular detection models: rider detection, vehicle detection, and relationship determination. Each module handles specific aspects independently, which manages system complexity while generating comprehensive risk profiles that include both rider and vehicle attributes.
Solution Approach 2:
The detection models are designed to be universal and adaptable to different micro-mobility vehicle types (bicycles, scooters, skateboards, etc.). By creating multi-functional models that can recognize various vehicle configurations and attributes, the system achieves comprehensive risk profiling without proportionally increasing complexity, as the same framework handles diverse vehicle types.
Data Source
AI summary
The systems and methods disclosed herein provide a risk prediction system that uses trained machine learning models to make predictions that a VRU will take a particular action. The system first receives, in a video stream, an image depicting a VRU operating a micro-mobility vehicle and extract the depictions from the image. The extraction process may be determined by bounding box classifiers trained to identify various VRUs and micro-mobility vehicles. The system feeds the extracted depictions to machine learning models and receives, as an output, risk profiles for the VRU and the micro-mobility vehicle. The risk profile may include data associated with the VRU/micro-mobility vehicle determined based on classifications of the VRU and the micro-mobility vehicles. The system may then generate a prediction that the VRU operating the micro-mobility vehicle will take a particular action based on the risk profile.