In order for the Pepper robot to grasp objects, a detection of those objects has to be performed. Object detection consists in detecting objects in the robot's surrounding, using its cameras and potentially other sensors. Solving this problem will allow robots to retrieve object's coordinates, that can later be used to reach and grasp them.

In recent years, impressive breakthroughs have been made towards the resolution of this problem , notably through the usage of deep learning. The goal of this internship is to implement, experiment and improve such an approach on our robots by detecting the position and orientation of daily life objects without modifying them.

Taking inspiration from the literature, the intern will evaluate and implement different solutions to process the data provided by the robot's sensors. After this evaluation, the intern will evaluate the best method to solve the problematic.

Location: Paris, in SoftBank Robotics Europe headquarters

Duration: 5+ months

Profile: Engineering Diploma / Master of Science / Master of Research (computer science or related)

Language: English is mandatory, French is optional

Programming language: Python, C/C++, ROS

Others : Git, Linux

Recommended: experience with machine learning / deep learning architectures, computer vision.

GitHub link is appreciated !