PathPulse.ai
Rerouting The Future
The problem PathPulse.ai solves
The Problem: Urban Mapping Isn’t Keeping Up
Cities are growing fast, but the tools we use to navigate them are stuck in the past. Most current mapping systems rely on static, outdated, or incomplete data—leading to:
- Inaccurate directions and poor route planning
- Delayed awareness of road hazards or traffic changes
- Limited insights into what’s really happening on the ground
- No way to predict issues before they happen
In short, the maps we use don’t reflect the real-time rhythm of the cities we live in.
Our Solution: PathPulse.ai – Smarter Mapping from the Ground Up
PathPulse.ai is building the next generation of urban mobility technology. By transforming raw video and metadata from everyday vehicle dashcams into live, intelligent, and decentralized maps, we’re creating a system that reflects the actual state of the city—right now.
What Makes Us Different
- Crowdsourced AI Mapping: Every driver becomes a data contributor. As vehicles move, real-time video and metadata are used to update maps far more frequently and accurately than traditional systems.
- Predictive AI with PulseNet: Our platform doesn’t just report what’s already happened—it predicts what’s about to. Using advanced pattern recognition, we can foresee traffic jams, accidents, and road issues before they even occur.
- Millimaps (A New Layer of Digital Maps): We’re not just making maps more accurate—we’re rethinking what maps are. Our “Millimaps” are living, digital twins of cities that reflect not just geography, but also real-time social, environmental, and economic activity.
- Community-Powered with $PULS Rewards: Users who contribute valuable data earn $PULS tokens. This incentive structure builds a vibrant, self-sustaining community of drivers and data providers who keep the system active and reliable.
- End-to-End Hardware + Software: We design both the hardware and the software, ensuring high-quality, AI-optimized data across different regions and driving conditions.
- Open to All: With the Pulse Scout app and our easy-to-use devices, anyone—from taxi drivers to delivery fleets—can participate and benefit from the system.
How It Works
- We combine real-time analytics, predictive AI, and big data
- Our system processes micro-level driving data to reveal patterns most systems miss
- Navigation maps update in real time as new data flows in
- City planners, businesses, and drivers get insights based on real-world conditions, not assumptions
Where It’s Used
- Detecting road hazards and incidents as they happen
- Optimizing delivery and transit routes in real time
- Helping cities plan and maintain infrastructure proactively
- Monitoring traffic and environmental patterns
- Rewarding users who contribute data through a token-based system
The Bigger Impact
PathPulse.ai is more than a mapping tool—it’s a real-time intelligence layer for cities. By blending predictive AI, community-powered data, and cutting-edge hardware, we’re enabling smarter, safer, and more sustainable mobility systems. This is the future of how cities move—and how we move through them.
Challenges I ran into
Problem 1: Multi-Device Camera Integration
Challenge:
Integrating dash cams from multiple car manufacturers proved difficult due to variations in camera hardware specifications, data transmission protocols, and lack of standardized interfacing methods.
Solution:
We addressed this issue using SPTP (Simple Picture Transfer Protocol) sniffing. By establishing a local server, we redirected the camera data streams to a centralized host through protocol sniffing and custom-built handlers. This abstraction allowed our application to interface seamlessly with a wide variety of dash cams, ensuring stable and consistent data ingestion across multiple devices.
Problem 2: Lack of Training Data for Model Development
Challenge:
There was a scarcity of relevant real-time annotated data required to train the AI models, particularly for identifying road features such as accident-prone zones, street lighting conditions, and object classification in diverse conditions.
Solution:
To overcome this, we launched a data collection campaign, recruiting a team of volunteers to gather real-world footage. Volunteers were instructed to capture specific image data and metadata such as GPS coordinates and timestamps. These curated datasets allowed us to develop a robust training pipeline for our AI model, incorporating edge-case scenarios and enhancing detection accuracy across diverse urban and rural landscapes.
Problem 3: Integration Limitations of YOLO with Telematics Hardware
Challenge:
The YOLO object detection model, while accurate, was computationally intensive and incompatible with the limited processing capacity of our in-vehicle telematics hardware.
Solution:
We developed a proprietary lightweight object detection model named PulseNet. PulseNet was engineered specifically for edge deployment—it is faster, consumes fewer resources, and delivers accurate and precise results under real-time constraints. This enabled seamless AI inference directly on the telematic device without compromising speed or accuracy.
Problem 4: Inability to Determine Contextual Object Positioning
Challenge:
Although PulseNet was effective in detecting objects, it lacked the ability to provide contextual spatial information, such as whether the object appeared on the left, right, or center of the road—critical for navigation and road behavior analysis.
Solution:
To resolve this, we introduced an extension module called PulseGen, designed to compute contextual positioning metadata. PulseGen works alongside PulseNet to analyze object placement relative to road geometry, thus enabling the system to identify lane-specific object locations. This improvement allowed for the generation of highly accurate and geo-contextualized mapping data.
Tracks Applied (1)
AI
Cheer Project
Cheering for a project means supporting a project you like with as little as 0.0025 ETH. Right now, you can Cheer using ETH on Arbitrum, Optimism and Base.