LogicTronix offers ADAS sensor fusion solutions for automotive OEMs and Tier-1 suppliers, built on Xilinx Versal or MPSoC FPGA platforms. These power-efficient heterogeneous and adaptive SoC platforms are well-suited for sensor fusion and machine learning acceleration.

A single device can process data from LiDAR, cameras, RADAR, and ultrasonic sensors, and perform ML workloads on fused sensor data.

LogicTronix is collaborating with a leading German automotive OEM and a North American Tier-1
automotive supplier to develop and deploy an advanced ADAS sensor fusion solution.

Feature of LogicTronix ADAS Solution

  1. Offering feature towards ADAS Level 2+ functionality
  2. Offering Sensor Fusion Capability with AI/ML + Algorithmic methods : with LIDAR, GMSL2/FPD Link III/IV Camera and Ultrasonic Sensors
  3. AI and Machine Learning–Accelerated Sensor Fusion Capability
    • Performs the Machine learning algorithms for feature extraction , processing streams and fusing of results/information.
    • We implement LIDAR based Point Cloud based 3D object detection with ML model, Object (Vehicle, Pedestrian, animal) detection with LiDAR and Vision data, recognition ML model, Lane detection/tracking algorithm.
    • Offers 30-45 FPS of inference across 4+ camera streams, with an independent LiDAR ML pipeline operating in under 20 ms, along with dedicated ultrasonic, 10G, and display pipelines.
    • All pipelines are synchronized for unified visualization with 3D rendering on a single chip.
  4. Offering parking aid feature with 8 number of Automotive grade Ultrasonic sensors
  5. Collision Avoidance system and early warning using LiDAR.
  6. The solution operates within a power range of 30W–40W.
  7. Supports AMD Versal platforms with a single-chip solution architecture, including the Versal AI Edge Series and Versal AI Edge Series Gen 2. The solution is available for evaluation on VEK280, VEK385, as well as select custom Versal platforms.

LogicTronix ADAS 2.2 – Sensor Fusion Solution on Versal [2026]

LogicTronix ADAS Sensor fusion solution leverage AMD-Xilinx Versal Adaptive SoC for Vision, LiDAR and other sensor data processing , running AI/ML on top of those and visualizing it. We optimize the algorithms and applications on Versal VEK280 with VE2802 AI Edge device and Versal AI Edge Series Gen 2, VEK385, as of so, the AI Engine in Versal used for LiDAR data processing , AIE-ML/AIE-MLv2 + PL for AI/ML and PS for managerial tasks.

LogicTronix ADAS Sensor Fusion with AMD Versal

Figure: LogicTronix Sensor Fusion with Versal – Block Diagram View

Versal AI Engine for High-Performance, Low-Power Sensor Fusion and Edge AI Computing

High-Performance Processing for Automotive Workloads

  • The Versal AI Engine is well-suited for LiDAR point cloud and radar data processing.
  • It delivers low-latency, high-throughput computing within a single Versal device.

Optimized Architecture for Parallel Computation

  • Equipped with dedicated vector and matrix processing capabilities.
  • Optimized for parallel mathematical operations used in sensor fusion, signal processing, and AI/ML workloads.
  • Enables improved computational efficiency and overall system throughput.

Deterministic and Hardened Compute Engine

  • Implemented as a hardened block within the Versal architecture.
  • Provides deterministic and predictable real-time performance.

Competitive Edge in Edge AI Computing

  • Delivers performance comparable to edge GPU platforms such as NVIDIA Orin.
  • Suitable for advanced automotive perception and decision-making systems.

Power-Efficient Automotive Solution

  • Offers significantly lower power consumption compared to edge GPU-based solutions.
  • Well-suited for power-sensitive automotive and embedded applications.

ADAS Sensor Fusion with Versal – Demo

We demonstrated our ADAS Sensor Fusion with AMD Versal Adaptive SoC – at AMD RTT , Kuala Lumpur, Malaysia at December, 2025.

LogicTronix ADAS 2.0 [2025]

LogicTronix ADAS 2.0’s features:

  1. 360 Surround View System for parking aid
  2. Collision Warning System with Vision ML and LiDAR ML model
  3. Drive-in ML model for object and signs detection
  4. Data Logging in 10G ethernet interface
  5. Perception and Navigation Stack – on development

Figure – ADAS 360 Surround View System with ZCU102 MPSoC FPGA with 4 RGB camera setup

LogicTronix ADAS 2.0 (2024)

Figure – LogicTronix ADAS 2.0 coming with LIDAR + 8 GMSL Camera Sensor + 8 UltraSound and AI/ML Capability

Watch LogicTronix ADAS Webinar [2022]

LogicTronix ADAS 1.0 [2022]

In MPSoC we are offering ADAS Sensor Fusion (Computer Vision+ LIDAR + UltraSound) and Machine Learning acceleration over those sensors input.

Figure – LogicTronix ADAS 1.0 Solution with 4 Camera + ML Implementation [ADAS- Driver Assistance]

The most suitable interface for ADAS solution are GMSL and FPD Link, our ADAS solution takes the GMSL 2 Camera link interface with more than 4 Camera and those camera are pre-processed.

And pre-processed stream is passed to Machine Learning Pipeline inside MPSoC FPGA.

Figure – LogicTronix ADAS prototyping and demo platform with 4 GMSL Camera-FMC-FPGA

Reference Tutorials by LogicTronix: ADAS [Automotive] Solutions

  1. ADAS Reference Design with Xilinx MPSoC FPGA and AI/ML – by LogicTronix [LRFD-031]: PDF Link
  2. ADAS tutorial at Hackster: Link

For more information about our ADAS solutions and features, please contact us by filling-up the form below.

Contact Us: