NOD G-1, a safety node for the road

The road should be able to say stop.

NOD G-1 is a local safety module in development. It is designed to detect defined road hazards and request a safe response when the primary autonomy stack or network connection is unavailable.

Sensing60 GHz radarsees range, speed and shape, never a face
Reflex< 50 mshold sent over a wire, our design target
Privacy0 bytesof personal data leave the pole
Power reserve20+ hr batteryLiFePO₄ backup power, design target
OriginU.S. engineeringTAA compliance by design

What is the NOD G-1?

One node on the pole could be the bridge.

NOD G-1 is a vehicle-local safety module in development. It is designed to detect defined road hazards and request a safe response when the primary autonomy stack or network connection is unavailable.

Roadside sensing can supply environmental context beyond a vehicle's own line of sight. The design keeps the simple wired reflex path separate from higher-level perception and does not assume control of a closed vehicle stack.

Inside, there are two paths. The wired safety output stays separate from the higher-level perception software.

  • Radar to reflex board to output, over copper. Under 50 ms is the target.
  • The compute module sends a continuous health packet. If it stops, the local reflex treats the compute path as unavailable and moves toward its defined safe state.
  • Each safety event is designed to create a compact record of what happened and why.
Sealed enclosurepole-mount concept with a weather-resistant enclosure Battery and chargingLiFePO₄ reserve with mains charging, design target Edge computeNVIDIA Jetson, classifies beside the reflex, never inside it Reflex boardsafety microcontroller and wired safety output 60 GHz radarsees range, speed and shape, never a face

The rules we keep

Four architectural anchors for design.

These shape how our system interacts with vehicles, infrastructure, and autonomy stacks built to evolve without locking us into rigid, brittle assumptions.

01

Local reflexes, secured communications.

The architecture is designed to execute defined safety triggers locally at the edge. Outbound coordination uses authenticated channels without forcing actuation onto an external vehicle.

02

Environmental truth over remote command.

The node acts as a local observer instead of trying to command a closed vehicle stack. It is designed to broadcast verified conditions and present clear physical cues so each vehicle can make its own safe response.

03

Perception data stays contextual and aggregate.

The design extracts only what matters for situational awareness, including obstacles, classifications, zones, and kinematic confidence. High-resolution raw tracking is intended to stay at the edge.

04

Every link has a health signal.

Each critical connection is designed to maintain a continuous liveness signal. If a data stream or communication link drops or degrades, the local safety logic can treat the silence as a fault condition.

Where it stands

The same node, three places it can stand.

One reflex, tuned to the place it is mounted. Pick a scene.

10 m20 m30 m OUT OF RANGE not tracked T-11 CYCLIST, 5.1 m/s T-07 PEDESTRIAN, 1.3 m/s T-03 VEHICLE, 9.4 m/s ADVISORY → T-03 HOLD REQUEST, crossing occupied NOD G-1 60 GHz mmWave, 120° FoV
A crossing, plan view120° field of view

At the crossing

The node is designed to watch a defined crosswalk and the lanes that feed it. When a person enters the protected zone, it can request a hold from an approved local interface and issue a warning to supported vehicles.

  • A relay output is part of the current traffic-cabinet integration concept.
  • The initial use cases are crossings, school zones, and streets shared with bikes and scooters.
  • The sensing output is intended to classify movement without creating an identifying image.
DIRECTION OF TRAVEL NOD G-1 one-direction 60 GHz coverage G-1 DETECTION → T-21 T-21 STOPPED, 0 m/s, LANE 2 T-19 TRUCK, 29 m/s VERIFIED SIGNAL ADVISORY → T-19, 1.2 km AHEADSLOW, stopped vehicle lane 2, merge left
A corridor, one direction of travelG-1 detects T-21 → warns T-19

On the corridor

The corridor concept focuses on one direction of travel. G-1 detects a stopped vehicle or debris to the right of the image, then sends a verified warning back to approaching traffic.

  • Warning distance depends on the deployment and supported communications link.
  • Rain, fog, spray, and darkness are required environmental test conditions.
  • V2X is used only where the road authority and supported vehicles provide it.
NOD G-1 battery reserve sensing continues GRID POWERdown CELLULAR / CLOUDdown LOCAL SAFETY LINKauthorized link, where configured APPROACHING VEHICLESforward announcement, live spatial awareness holds Blackout scenariowhat still works when the infrastructure around the node fails
A blackout, schematicWhat still works

When the grid goes down

The design uses battery reserve so local sensing and event records do not depend on cloud access. External communication depends on the authorized links configured for that deployment.

  • Battery reserve supports local sensing when mains power fails.
  • Cloud-independent operation keeps the local safety path available.
  • Vehicle or agency messages use only approved radios, spectrum, and operating authority.

Why 60 GHz

Physics draws the boundary.

We chose 60 GHz on purpose. A wide sweep gives fine detail, and the air itself keeps the signal close. Drag the slider to see what the sweep does.

7 GHzSweep bandwidth B
2.1 cmRange resolution Δr
RADAR RETURN AGAINST RANGE persondebris, 20 cm away

The return shows two objects. A person and the debris beside them stay separate.

Δr = c2B = 3 × 108 m/s2 × 7 × 109 Hz ≈ 2.1 cmγO₂(60 GHz) ≈ 15 dB/km

Range resolution from sweep bandwidth, and oxygen absorption at sea level.

A wide sweep

Fine enough to tell a person from debris

The 57 to 64 GHz band allows up to 7 GHz of continuous sweep. Detail improves as the sweep widens, so the node can resolve about 2 cm. That is enough to keep a person standing beside a car separate from the car.

Air that absorbs

A bubble with a natural edge

Oxygen absorbs energy near 60 GHz. That attenuation helps keep short-range sensing local. Deployment and radio operation remain subject to the applicable regional rules and validated hardware.

A wider sweep sees finer detail Δr =c2B 020406080100 GHz 24 GHz ISMB = 250 MHzΔr ≈ 60 cm 57 to 64 GHzB = 7 GHzΔr ≈ 2 cmNOD G-1 76 to 81 GHzB = 5 GHzΔr ≈ 3 cm Oxygen absorbs 60 GHz, so the signal stays close specific attenuation, sea level dB / km 0.010.1110100 020406080100 GHz ≈ 15 dB/km at 60 GHz the oxygen resonance peak 24 GHz ≈ 0.2 77 GHz ≈ 0.4 The return fades past the crossing, so the node reads its own corner and ignores the rest.
Sweep bandwidth and oxygen absorptionAttenuation at sea level

The cars already on the road

Modern autonomy relies on perception. We provide the ground truth.

V2X lets vehicles perceive environmental context beyond line-of-sight. While next-generation platforms adopt connected channels and autonomous stacks guard their perimeters, existing fleets need a simple bridge. We provide the environmental truth vehicles need to make safe stops, from diagnostic-port fleet adapters to roadside perception broadcasts.

Today's vehicleDiagnostic / Service Port

Standardized diagnostic access is common on U.S. light-duty vehicles from model year 1996. Available signals and permitted actions vary by vehicle and fleet interface.

In developmentPrometheus bridge

Designed to read available vehicle dynamics, authenticate safety data, and connect in-cab awareness to local road conditions.

The radioV2X, 5.9 GHz

A standardized direct link for cooperative perception where regional rules and vehicle support allow it. Bluetooth tool mesh is separate and is not V2X.

On the poleNOD G-1

Designed to perceive a crossing, verify the local condition, and issue a safety cue for approaching traffic.

Vehicle support depends on model, region, OEM implementation, and approved interface

What people are saying

The intelligence belongs on the road.

“Reactive AV safety built on machine learning is unsustainable. Proactive safety must be embedded in road infrastructure.”

Ricky Kwok, as featured in Frenus, Best of LinkedIn, Future Mobility and Market Evolution, CW 23/24

“Vehicles from different manufacturers need to communicate not only with each other but with infrastructure and pedestrians.”

Federal Highway Administration, cooperative driving automation

“Autonomy will never scale commercially as long as liability is trapped inside a vehicle-isolated stack.”

Dr. Priyalatha, Lead, Automotive Communications and Networking

“An ingenious invention that promises to advance AV technology and the overall safety of everyone.”

Julian Wong, writer for Eurekascoop, on the Sovereign Spine
Frenus Best of LinkedIn badge

Featured, June 15, 2026

Ricky Kwok's post was selected as a strategic insight in Frenus, Best of LinkedIn, Future Mobility and Market Evolution, CW 23/24.

Built on Texas Instruments 60 GHz radar and NVIDIA Jetson edge compute. NVIDIA Inception member since June 2026.

Talk to us

Tell us what you are working on.

A city crossing, a fleet, a fireground, a port. If people and machines share the space, we would like to hear about it. The founding team reads every message.

Goes straight to the founding team.