RaZIR®

Double your radar’s angular resolution through software alone — no hardware changes required. RaZIR improves the separation of closely-spaced objects on any imaging radar, giving your perception stack the precision it needs to distinguish targets.

DRIVER’S VIEW · TWO VEHICLES AHEAD
WITHOUT RaZIRRADAR POINT CLOUD1 detectionLATERAL · Y (m)RANGE · X (m)
WITH RaZIRRADAR POINT CLOUD2 detectionsLATERAL · Y (m)RANGE · X (m)

Applications

Built for sectors that demand precise target separation.

Automotive

Separate closely-spaced vehicles, pedestrians and parked cars for AEB, FCW and lane-change assist.

Industrial

Resolve people, machinery and infrastructure in cluttered yard, mining and construction sites.

Agricultural

Detect workers and animals through dense crop clutter where vision-only sensing fails.

Drone

Separate and track multiple small airborne targets for counter-UAS and airspace monitoring.

How it works

A three-stage pipeline that refines radar output without touching hardware.

STEP 01

Detect areas of interest

A fast, low-compute scan identifies regions containing potential targets. Empty space is discarded early so refinement effort goes only where it matters.

AREA OF INTEREST empty · skip empty · skip

STEP 02

Zoom in and resolve

Proprietary resolution-enhancement algorithms refine the angular estimates inside each region, separating returns that FFT-based processing alone leaves merged.

Iteration 1 · FFT — two targets blur into one peak Iteration 5 · refined — peaks resolved as distinct

STEP 03

Output enhanced point cloud

An enhanced point cloud is emitted to downstream perception, directly improving object detection, classification and tracking with no changes to your inference pipeline.

FFT Single return RaZIR Two distinct returns 2× RESOLUTION

Featured deployment

When cameras and LiDAR fail, radar with RaZIR doesn’t.

A global agricultural OEM deployed imaging radar running RaZIR-enhanced resolution and maintained clear detection of farm workers walking through dense corn crops — where cameras and LiDAR failed due to visual occlusion. The radar penetrated the vegetation and resolved individual personnel from surrounding clutter, enabling safety-critical decisions that vision-only sensing couldn’t.

  • Workers detected through dense vegetation
  • Personnel resolved from crop clutter
  • Camera and LiDAR coverage gap closed by radar

Variants

Three flavours, sized to your compute envelope.

The same resolution-enhancement IP runs on anything from a constrained automotive SoC to an unconstrained GPU.

MAXIMUM QUALITY

Deep RaZIR

For unconstrained processing — GPUs and high-end SoCs. Highest angular precision and the deepest object separation in the family.

Target hardware: High-end automotive SoCs and GPUs

STANDARD

RaZIR

The mainstream variant. Balanced for a wide range of SoCs and mid-range GPUs — the default for most automotive and industrial deployments.

Target hardware: Mainstream automotive SoCs

MAXIMUM EFFICIENCY

RaZIR Lite

For processing-limited applications — small SoCs and edge nodes. Retains the core resolution gain at a fraction of the compute budget.

Target hardware: Constrained SoCs, edge / embedded nodes

Integration

Works with your hardware

RaZIR is architecture-agnostic. It can be licensed into non-Provizio processing chains and runs across the compute and radar platforms your stack already uses.

  • Production automotive SoCs
  • GPU architectures
  • Generic CPU and DSP implementations
  • Any imaging radar antenna design

Explore the stack

Pair RaZIR with the rest of Provizio’s perception modules.

MIMSO® Antenna

Patented software-defined sparse array antenna technique that increases resolution while keeping side-lobes low.

5D Perception®

Provizio’s complete AI perception stack — software for end-to-end radar scene understanding.

NoiseReject

Enhanced signal clarity in reflective environments by suppressing noise and ghost detections.

DEEP Radar Detector

Next-generation ML-powered CFAR alternative for advanced signal processing, object detection and classification.

Questions, answered

RaZIR FAQ

RaZIR is Provizio’s software module for radar angular resolution enhancement. It applies proprietary zoom-in algorithms to regions of a radar scan that contain objects of interest, delivering up to 2× improvement in angular resolution without changes to the radar front-end.

Classical methods like MUSIC and ESPRIT apply uniformly across the scene, which is expensive in real time. RaZIR is region-selective — it runs heavy refinement only inside detected areas of interest, keeping compute load predictable on production automotive and industrial hardware.

Any imaging radar. RaZIR is a software module licensed into a customer’s existing processing chain — it works with any antenna and front-end architecture.

It depends on scene complexity and which variant is deployed — Deep RaZIR targets unconstrained processors, standard RaZIR is sized for mainstream SoCs, and RaZIR Lite is for constrained edge nodes. Reach out for more info.

Ready to evaluate RaZIR?

Talk to us to discuss how RaZIR can fit into your radar stack.

Contact us Learn more