AI Technologies
The AI-based Object Recognition (OR)
AI-based systems enable automatic object detection and collision-free navigation for EvoLogics autonomous vehicles.
Optimized for the limited resources of uncrewed platforms, all processing is performed onboard and in real time, providing responsive awareness for safe and reliable missions.
With adaptive AI, operators benefit from faster, more accurate surveys, reduced human intervention, and enhanced mission safety, while maximizing vehicle performance.
Extensively tested with our selected partners, the Object recognition system is available as an add-on option for all EvoLogics commercial vehicles - including the Sonobot 5 USV and the Quadroin AUV.
AI Technologies
Object recognition
The AI-based Object Recognition (OR) system is an additional module that runs directly onboard the autonomous vehicle and analyzes raw side-scan sonar and video camera feed. Objects of interest are detected and visually highlighted in the operator’s control software - all in real time during the mission.
A cloud-based ecosystem supports the OR system with regular updates and new detectable object classes. To keep recognition capabilities aligned with the evolving operational needs, the cloud service welcomes user contributions, enabling dataset uploads and requests to train the system for objects contained in the scans.
Addressing the challenge
Recognizing objects in side scan sonar visuals traditionally requires extensive training and experience, as operators must notice and interpret subtle cues in contours and shadows. One often needs to perform multiple passes to collect data from different angles. Thus, a survey can be very time-consuming and tiring for the operator who is often working under pressure, outdoors, and in poor weather.
To address the challenges of locating and identifying submerged assets, EvoLogics developed a system for automatic object recognition in data collected by our autonomous vehicles.
Our AI-based solution uses neural network algorithms designed to process side-scan sonar and video data, handle large data volumes in real time, and run efficiently on dedicated embedded hardware - since cloud computing is not available in open waters. Our pre-trained models cover a wide range of objects relevant to key customer groups. The system is ready to use out of the box, and additional training is available to expand its object recognition capabilities.
Extensively tested with our selected partners, the Object recognition system is available as an add-on option for all EvoLogics commercial vehicles - including the Sonobot 5 USV and the Quadroin AUV.
Object Recognition - Key features
-
Real-time object recognition
Raw side-scan sonar and video data is analyzed on the fly onboard the vehicle
Optimized performance - detecting objects in less than a second
Advanced neural network algorithms process high-volume sonar and video data on dedicated embedded hardware
-
Expanding range of object classes
Use our existing object library
Leverage a cloud-based service platform that regularly adds detectable object classes
Add your own objects - perform scans and submit your data for AI training
-
Cloud-based services
Regular updates for the on-board software module
Create new object classes by submitting your scans for analysis and AI training
Choose to share your object classes within the OR cloud ecosystem or restrict them for private use
Military-grade data security ensures full protection of all uploaded datasets
-
Side-scan sonar and video feed integration
Works seamlessly with raw side-scan sonar output and video feeds from onboard cameras.
The same hardware and software modules efficiently handle both data types.
-
Compatible with all EvoLogics vehicles
Sonobot 5 USV Uncrewed autonomous vehicles
Quadroin AUV, Poggy AUV
Available as a hardware and software upgrade for existing Sonobot USVs
-
Expanding range of objects
Use our existing objects library
Cloud-based service to train the system for new objects
Add your own objects - perform scans and send your data for AI training
Obstacle Avoidance System
Ensuring safe navigation in dynamic environments is critical for autonomous marine vehicles. The AI-powered obstacle avoidance system enables our USVs and AUVs to detect and avoid obstacles in real time, maintaining mission efficiency while minimizing risk.
Obstacle Avoidance System – Key Features
Depending on the vehicle's configuration, the system integrates data from an underwater front-looking sonar, an overwater stereo camera, and a LiDAR sensor to continuously monitor the surroundings and trigger automatic maneuvers that prevent collisions.
Built on advanced neural network algorithms, the system processes sensor data onboard, generating actionable navigation decisions to protect both the vehicle and the operational environment.
-
Continuously monitors the vehicle’s surroundings using the forward-looking sonar, stereo camera, and LiDAR data - depending on the vehicle’s configuration.
Advanced neural network algorithms process high-volume sensor data on dedicated embedded hardware to identify potential obstacles beneath or above the surface.
-
Embedded neural networks analyze sensor data onboard the vehicle in real time.
Generates actionable navigation adjustments to avoid obstacles while maintaining mission trajectory.
-
Works natively with Sonobot 5 USV and our AUVs, using the same AI framework for surface and underwater platforms.
Depending on the vehicle configuration, it combines sensor inputs from forward-looking sonar, a stereo camera (USV), and LiDAR (USV) into a unified obstacle map for precise navigation and automated avoidance maneuvers.
Fully compatible with the vehicle’s autopilot and mission control systems, ensuring effective operation without operator intervention.
-
Forward-looking sonar for underwater obstacle detection - available for Sonobot 5 USV and AUVs.
Stereo camera and/or LiDAR for depth perception and obstacle detection above the water - available for Sonobot 5 USV.
