Technology Fundamentals

Product Operation Flow

The PathPulse system employs a robust, multi-phase approach to data-driven navigation, integrating real-time data collection, synchronised processing, and AI-based decision-making to ensure responsive and adaptive route guidance.

Data Collection Phase:

• Any Dash or Phone Camera captures continuous visual information data of the environment

• The IMU sensors gather precise location data (GPS coordinates and timestamps)

• Both data streams are fused together for synchronised processing

Storage & Processing Phase:

• The frame-by-frame collected data is combined and stored EMMC memory for stable storage

• The PathPulse dash camera captures 15 frames per second, balancing efficient data collection with high-quality video, unlike many dash cameras with higher frame rates.

• Once the user's phone is connected to the PathPulse application, the data is uploaded.

• The advanced processor handles initial data processing.

Decision-Making Phase:

• Processed data is sent to the cloud for advanced analysis

• AI algorithms analyse the environment and current position

• The system makes navigation decisions based on:

• • Current position and destination

  • Detected obstacles or path conditions

  • Real-time environmental changes

  • And much more…

Example Action Loop:

• If the path is clear, The system continues on the current route

• If obstacles are detected: The system recalculates path and feeds new data back to the processor

• This creates a continuous feedback loop for real-time navigation

Advanced Object Detection

We employ PulseNet, a state-of-the-art object detection model, to analyse visual data from phone cameras, dashcameras and traffic cameras. PulseNet’s architecture offers an optimal balance of speed and accuracy, which is crucial for identifying road hazards, traffic signs, and potential accidents in real-time.

Key advantages of our PulseNet implementation:

• Real-time processing of multiple object classes simultaneously

• High accuracy in detecting small objects like potholes or distant traffic signs

• Compact model size, ideal for edge deployment

Edge Computing with a Powerful Chipset

PathPulse.ai leverages a chipset, which is a powerful yet energy-efficient AI accelerator, to process data at the edge. This approach significantly reduces latency and enhances system reliability, especially in areas with poor connectivity.

The Neural Processing Unit (NPU) is specifically optimised for AI workloads, allowing us to run complex inference tasks with minimal power consumption—a critical factor for mobile and in-vehicle systems.

Optimised Model Deployment

To maximise performance on the AI accelerator, we utilise a specialised deployment pipeline:

1. Train the PulseNet

2. Convert the model to ONNX (Open Neural Network Exchange) format for portability

3. Use the chips (name is confidential) toolkit to optimise the model for hardware

4. Deploy the optimised model on devices, leveraging the NPU for accelerated inference

This process ensures that PathPulse.ai can deliver real-time insights even on resource-constrained edge devices.

Data Fusion and AI-Driven Analytics

Beyond visual data, PathPulse.ai integrates information from various sources:

• IoT sensors for environmental monitoring

• Telematics data for vehicle behaviour analysis

• Social media detected incidents for updates

Our proprietary AI algorithms sift through this diverse data stream, identifying patterns, predicting traffic flows, and generating actionable insights for users and city planners alike.

Blockchain for Data Integrity

We employ blockchain technology to ensure the reliability of community-contributed data. This decentralised approach guarantees data integrity and enables a transparent reward system for active community members.

5G Integration

As 5G networks continue to expand, PathPulse.ai is positioned to leverage high-speed, low-latency connections. This will enable even more responsive and detailed urban mapping, paving the way for advanced features like real-time augmented reality navigation.

By combining these technologies, PathPulse.ai creates a robust, scalable platform capable of processing vast amounts of urban data in real time. The result is a living, breathing digital twin of the city, offering unprecedented insights and enabling smarter, more responsive urban environments.