Perception for humanoid robotics
We offer Master theses/Guided Research topics in the field of computer vision and robotics, in collaboration with the Chair of Applied Mechanics
The project topics concern the application of computer vision and 3D perception techniques to humanoid robotics. In particular, they focus on one of the following two tasks:
- Object detection and tracking for the humanoid Lola: Lola is the humanoid being developed at the Chair of Applied Mechanics (as shown in this demonstrative video). The main task of this project is to develop an object detection and tracking module that runs in parallel with the existing threads on Lola's operative system and detect a specific object class (e.g., a chair or a box) in the scene while keeping track of it. The goal is to allow the humanoid to follow the object moving around a scene through its navigation and locomotion modules.
- SLAM for the humanoid Lola: the idea of this project is to integrate a Simultaneous Localization and Mapping (SLAM) algorithm as an additional perception module within Lola's operative system. Through SLAM, we aim to reduce the drift currently present in the robot's kinematic system, as well as be able to obtain a 3D reconstruction of the environment necessary for the estimation of surfaces (in particular, walls) that are present in the surrounding environment.
The goal of this project is to handle the two main issues of mounting SLAM system on a robot:
(1) the vibration caused by the locomotion of robots which may lead to tracking-loss of the robot's current position, and
(2) the feature-less regions which do not have enough information for SLAM to perform localization.
More information about the LOLA project please refer to this page
If you are interested in any of these topics, please contact us via e-mail:
| ProjectForm |
| Title: || Perception for humanoid robotics |
| Abstract: || This thesis studies unsupervised monocular depth prediction problem. Most of existing unsupervised depth prediction algorithms are developed under outdoor scenarios, while depth prediction in the indoor environment has long been ignored. Therefore this work focus on filling the gap by first evaluating existing architecture in the indoor environments and then improving the current design of architecture by solving observed issues in the first step. After an extensive study and experiment in the current methods, an issue has been found that existing architecture cannot learn depth freely by the reason of the poor performance of the pose estimation network, which is a side network usually being trained together with depth prediction network. Unlike typical outdoor training sequence, such as Kitti dataset, indoor environment consist of more arbitrary camera movement and short baseline consecutive images which contribute poor training to the pose network. To address this issue, we propose two methods: First, we design a reconstruction loss function to provide extra constraint to the estimated pose and sharpen the predicted disparity map. Second, a novel neural network architecture is proposed to predict accurate 6-DOF pose. Our pose network combines the advantage of FlowNet2? and PoseNet? which makes it able to learn to predict correct poses with relatively short baseline and arbitrary rotation training images. Apart from the above two methods, we use an ensemble and a flipping training techniques along with a median filter on the output disparity map, resulting outperformance of the current state-of-the-art unsupervised learning approaches. |
| Student: || |
| Director: || Federico Tombari |
| Supervisor: || Shun-Cheng Wu |
| Type: || Master Thesis |
| Area: || Computer Vision |
| Status: || draft |
| Start: || |
| Finish: || |
| Thesis (optional): || |
| Picture: || |