StradVision and AMD unveiled groundbreaking multi-camera perception technology at CES 2026, showcasing SVNet MultiVision software running on AMD’s Versal AI Edge Series Gen 2 platform. The demo delivers up to 185 INT8 TOPS compute power, enabling real-time, synchronized processing of multiple camera streams for Level 3 autonomous driving.
Table of Contents
StradVision SVNet MultiVision: Production-Ready Perception
The demonstration uses the AMD Versal AI Edge Series Gen 2 VEK385 Evaluation Kit as a technical reference, highlighting how production-proven perception software scales across heterogeneous compute resources. The platform processes simultaneous camera streams for advanced ADAS and automated driving use cases with high-throughput performance.
This CES showcase builds on a multi-year collaboration announced in November 2025, following initial joint demonstrations at CES 2025. The partnership targets software-defined vehicles and scalable perception architectures for OEMs and Tier-1 suppliers evaluating next-generation autonomous systems.
| Specification | Details |
|---|---|
| Software | SVNet MultiVision perception stack |
| Hardware | AMD Versal AI Edge Gen 2 VEK385 |
| Compute Power | Up to 185 INT8 TOPS / 370 MX6 TFLOPS |
| Capabilities | Multi-camera sync, object detection, lane recognition |
| Autonomy Levels | Level 2 ADAS to Level 3/4 prototypes |
| AI Engines | AIE-ML v2 (quantized inference, near-FP16 accuracy) |
| Target Market | Software-defined vehicles, centralized compute |
| CES Booth | AMD W223, LVCC West Hall Level 2 |
Bridging L2 to L3 Autonomy
The collaboration addresses a critical automaker challenge: transitioning from Level 2 assistance to Level 3 autonomy without redesigning entire electronic architectures. The solution enables L2 systems with rich object, lane, and free-space detection, plus L3 prototypes with high-precision perception for real-time environmental modeling.
The AMD Versal platform leverages integrated AIE-ML v2 engines to run quantized inference with near-FP16 accuracy while cutting power and latency—a critical balance for edge deployment inside vehicles where thermal management and energy efficiency determine viability.
StradVision’s MultiVision software optimizes advanced perception features including pedestrian detection, environmental understanding, and free-space recognition. These capabilities enhance both performance and thermal efficiency in autonomous driving systems, meeting stringent automotive safety requirements.
For comprehensive autonomous driving technology analysis and automotive AI reviews, explore TechnoSports. Learn more about StradVision‘s perception software and AMD’s automotive platforms.
The combined solution accelerates time-to-market through pre-integrated, production-ready AI perception while reducing system complexity and power consumption. With ISO 26262 functional safety certification and TISAX AL3 information security standards, StradVision delivers validated capabilities for real-world deployment.
This demonstration marks significant progress in the expanding AMD automotive ecosystem, building foundations for AI perception that scales intelligently across software-defined vehicle architectures.
FAQs
What compute power does AMD Versal AI Edge Gen 2 deliver?
The platform provides up to 185 INT8 TOPS or 370 MX6 TFLOPS in automotive-qualified packaging.
What autonomy levels does the StradVision-AMD solution support?
It enables Level 2 ADAS systems and supports prototyping for Level 3/4 autonomous driving applications.
