NuroBuckle | Devpost

Inspiration

Inspiration came from the alarming reality that driver fatigue is a major cause of truck crashes—contributing to 13% of large truck accidents and up to 40% of heavy truck crashes. Every year, fatigue-related incidents lead to 71,000 injuries, 800 deaths, and over $100 billion in economic losses. Existing solutions like camera monitoring and time tracking are reactive, often inaccurate, and largely ignored by drivers. We saw an opportunity—and a responsibility—to build something better.

What it does

EEG-Based Fatigue Detection: NuroBuckle utilizes the Emotiv Insight 5-Channel wireless EEG headset to monitor the driver’s brainwaves and thus fatigue in real-time. This allows us to detect drowsiness, attention drops, and microsleep indicators that can’t be seen through traditional methods like eye-tracking or facial analysis.
Smart Route Adjustment: Based on the driver’s fatigue level, NuroBuckle can dynamically adjust the route, adding rest stops, gas stations, or restaurant detours to ensure the driver gets the necessary breaks.
Proactive Company Alert System: Using the Twilio MCP server, NuroBuckle sends real-time SMS notifications to both the driver and fleet manager whenever fatigue is detected, ensuring timely intervention and communication with dispatch.

How we built it

To build NuroBuckle, we integrated a range of technologies for real-time monitoring, data processing, and fleet communication:

EEG-Based Fatigue Detection: We used the Emotiv Insight 5-channel wireless EEG headset to capture brainwave data from the driver in real time. Configuring raw EEG signals to cognitive states such as attention, drowsiness, and microsleep. By processing these signals, we can detect early signs of fatigue that traditional methods like facial analysis or eye-tracking miss.

Signal Processing with Reinforcement Learning: The raw EEG data can be noisy, with signals from unrelated brain activities interfering with fatigue detection. To address this, we employed Cortex API for their reinforcement learning. This technique helps the system focus on relevant EEG channels related to fatigue and ignore irrelevant data, enhancing the accuracy and robustness of our fatigue detection algorithm.

Real-Time Route Optimization: We integrated Google Maps’ Real-Time Traffic API to dynamically adjust routes based on the driver's fatigue level. When the system detects that the driver is fatigued, it automatically suggests nearby rest stops, gas stations, or scenic detours. The API allows us to account for real-time traffic conditions and suggest the fastest and safest options for the driver to take a break.

Communication and Alerts: Using Twilio’s MCP Server, we built a robust communication layer that sends real-time alerts. When the system detects fatigue, Twilio SMS notifications are automatically sent to both the driver and fleet manager, ensuring that intervention can happen immediately. The driver is notified to take a break, and the fleet manager is alerted to the situation, allowing for further actions such as re-routing or adjusting the schedule.

Real-Time Fleet Management Dashboard: We used Streamlit to create a user-friendly dashboard for fleet managers. This real-time web interface allows managers to monitor the status of drivers, view their fatigue levels, and receive instant notifications. The dashboard provides an intuitive interface for efficient decision-making, ensuring that fleet managers can act quickly when fatigue is detected.

Model Context Protocol (MCP): The Model Context Protocol (MCP) is the backbone of this integration. MCP allows all components—EEG data processing, route optimization, and communication systems—to work in real time and exchange data seamlessly. It ensures that the system can react immediately, rather than relying on disconnected data sources or slow responses. This is what enables the proactive, real-time actions that make NuroBuckle unique.

By combining these technologies, we created a system that not only detects fatigue but also takes immediate, context-aware action to ensure driver safety.

Challenges we ran into

One of the major challenges was integrating EEG signal processing with real-time route optimization and fleet communication. The noise in the raw EEG signals were useless and we needed to find ways to use reinforcement learning to isolate fatigue signals accurately. Additionally, syncing multiple components, such as EEG data processing, route adjustments, and communication, in real time was complex but essential for ensuring timely interventions.

Accomplishments that we're proud of

The coordination of fatigue detection, real-time route optimization, and fleet communication is only possible with the Model Context Protocol. Unlike traditional systems, which only issue alerts, MCP enables proactive actions. When fatigue is detected, NuroBuckle can automatically reroute the vehicle to a safe stop, notify the company, and send SMS alerts—all in real time. Previous systems face bottlenecks in their integration of data sources. For example, Volvo and Samsara rely on separate systems for monitoring, alerting, and communication, which creates delays and inefficiencies. MCP eliminates these issues by standardizing communication between EEG data, route optimization, and fleet communication, allowing for immediate, context-aware interventions and a much more responsive system.

What we learned

What's next for NuroBuckle

We plan to refine our platform with direct integration with trucks, allowing for a deeper connection with vehicle telematics and a more comprehensive understanding of driver fatigue in relation to vehicle data. By continuing to refine this system, we hope to improve road safety by not only detecting fatigue but also proactively managing it in real time. NuroBuckle represents a step forward in the evolution of driver safety systems, ensuring that fatigue is no longer a silent killer on the road.