ecoRobotix develops robot weeders for ecological removal of weeds in several row crops. Our solar powered and lightweight weeding robot autonomously navigates the field and applies herbicide precisely and uniquely on the weeds that are detected. The weed detection from images is done using a convolutional deep learning network that performs pixel-level plant classification and that is trained on tens of thousands of images. Compared to standard methods of chemical weeding our system allows using 90% less herbicide, improving the crop yield and reducing the environmental impact and weeding costs. ecoRobotix also develops a scouting robot that performs autonomous navigation in the field and allows users to integrate the sensor of their choice for various applications of scouting and phenotyping.