Live Edge Inference Active

Ride Safer.
Feel the Threat.

Bridging the safety gap for two-wheeler riders through AI-powered haptic intelligence. AI-powered perception. Haptic intelligence. No distractions.

See How It Works Tech Stack
<50msInference Latency
3-DoFHaptic Vectors
Edge AINo Cloud Needed
1.35M Annual Road Deaths Worldwide
29% are two-wheeler riders
<50ms Edge Inference Latency
Real-time threat detection
3-DoF Haptic Directional Vectors
Left · Right · Rear
98.2% Collision Prediction Accuracy
On benchmark dataset
System Pipeline

From pixels
to haptic pulse

Four stages. One continuous loop. Running at the edge, milliseconds from danger.

01
Monocular Vision Capture

Rear-mounted wide-angle camera streams live video to the onboard edge computer at high frame rates.

02
AI Perception Stack

Simultaneous object detection, monocular depth estimation, and lane segmentation run in parallel on the edge.

03
Collision Probability Engine

Fused perception outputs feed a real-time risk model that continuously scores approaching vehicle threat levels.

04
Haptic Alert Delivery

Wireless signal dispatches directional and intensity-graded vibrations to the rider's smart gloves instantly.

Risk Model

Collision probability
in real time

The fusion model scores approaching vehicles continuously and dispatches haptic alerts proportional to threat level and direction.

Threat Scoring — Live Simulation
Vehicle A
18%
Vehicle B
54%
Vehicle C
87%
Haptic Glove Activation
LEFT GLOVE
THREAT DETECTED
RIGHT GLOVE
CLEAR
Monocular Depth Fusion

Depth cues extracted from a single RGB stream using self-supervised neural networks, eliminating the need for stereo rigs or LiDAR.

Bayesian Risk Estimation

Object trajectories and time-to-collision are combined in a probabilistic model that tolerates sensor noise and partial occlusion.

Directional Haptic Encoding

Threats are spatially decomposed into left, right, and rear channels, mapped to actuator intensity levels from 0–255.

Technology

The full stack

Every layer designed for affordability without compromising accuracy. AI does the heavy lifting so the hardware doesn't have to.

Monocular Camera Affordable, single-sensor setup — no LiDAR required
Edge AI Compute On-device inference, no cloud latency dependency
Depth Estimation Self-supervised monocular depth from a single RGB frame
Object Detection Custom-trained model optimised for rear-view vehicle classes
Lane Segmentation Context-aware lane occupancy and trajectory prediction
Haptic Gloves Multi-axis vibrotactile actuators with variable intensity
Wireless Protocol Ultra-low-latency BLE 5.3 glove-to-system link
Risk Fusion Model Bayesian collision probability from multi-stream inputs
Inspiration

Proven in aviation.
Built for the road.

Haptic feedback cuts pilot reaction time by 40% compared to audio alone. We are bringing that same principle to every rider.
— Adapted from aviation tactile alert research
01 The aviation industry's stick shaker system uses tactile feedback to alert pilots to stall conditions — an approach proven over decades of flight safety.
02 Unlike audio or visual alerts, haptic feedback bypasses conscious attention and reaches the rider even in noisy, high-speed conditions.
03 Directional encoding through the gloves mirrors proprioceptive navigation — the same instinct humans use to sense their own body in space.

Safety for
every rider.

Join our early access programme. Be among the first riders and partners to experience intelligent haptic safety.

Request Early Access