Driving the future

This capstone project, undertaken in collaboration with a team at Carnegie Mellon University’s Heinz College, focuses on building a predictive system that leverages telematics data and machine learning to forecast vehicle wear and tear before production. By integrating driving behaviors, environmental conditions, and performance metrics into 3D visualizations, the approach enables a virtual evaluation of vehicle reliability under a variety of real-world scenarios—reducing the need for expensive physical prototyping.

‍

The system operates through a layered process of data collection, feature engineering, and predictive modeling, culminating in clear, data-driven insights for vehicle design and maintenance. Utilizing tools such as Python, Stable Diffusion XL, genmo/mochi-1-preview, and Dash Plotly, the team processes high-volume telematics inputs (e.g., braking events, weather, incidents) to identify patterns and anticipate potential issues in vehicle components.

‍

A key advantage of this solution is its capacity to streamline the design and testing phase. By providing reliable performance forecasts, the project helps automakers and researchers minimize trial-and-error experimentation, cut costs, and accelerate time-to-market. This method also opens the door to iterative improvements in safety and functionality by showcasing how various design choices might fare under different conditions.

‍

Extensive modeling and visualization techniques—supported by machine learning frameworks such as Random Forests, PCA, and Generative AI Models (LLMs)—allow the team to transform raw telematics data into actionable recommendations. Engineers and stakeholders can then refine their prototypes or maintenance strategies based on quantified insights, enabling a more informed, evidence-based workflow.

‍

In summary, this Heinz College capstone demonstrates how telematics and machine learning can reshape automotive R&D by delivering predictive analyses that reduce reliance on physical testing. By harnessing advanced data pipelines and intuitive 3D visualizations, the project paves the way for safer, more efficient vehicle designs, ultimately driving innovation in the automotive industry.