UNMANNED OFF-ROAD VEHICLES
OUR customer is a technology player in agriculture industry. They wanted to do a feAsibility study on using un-manned off-road vehicles used in the in dustry to perform various jobs including tracking of fields, plats and weeds.
REQUIREMENTS
CREATING AI TRAINING TOOLS
Create data sets and training neural networks models for the specific tasks. Had to be easy to use for junior engineers.
TrACKING OBjects
Tracking both static and moving obstacles in real time including road signs and road irregularities .
Specialized agricultiral tracking
Tracking of agricultural terrains, including identifying of planting lanes and weeds.
ai training
Our specialists developed a web service for the client to manage datasets and training neural network models including annotation tools, access to ready made datasets, module for training models and analyzing metrics. The AI training solution provided the integration with Tensorflow, Caffe and Keras.
SOLUTION
Tartan Technology developed solution algorithms based on the latest advances in computer vision technology.
Obstacle detection and terrain assessment included:
Visual odometry
Neural network
Segmentation, Detection and Classification
Arable row tracking
Detection and tracking markers
Software API for specialized custom applications