Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Continuous online learning for intelligent autonomous robots
Most tools in robotics remain static after initial training. However, robots are deployed in varying conditions for which the initially trained modules are often sub-optimal. In this project we aim to enable robots to continuously learn and adapt to new circumstances on the fly.
Keywords: Online learning, Artificial Intelligence, Continuous Learning, Robot Localization, Robot perception, Deep Learning, Autonomous Robots, Construction Robotics
Capabilities of robotic perception systems have greatly improved since the advent of modern deep-learning based algorithms. One major drawback however, are their degraded performance under changing conditions or in new environments, preventing us from creating truly adaptive and reliable robots.
In this project, we will enable our robots to continuously adapt to new environments given cameras, LiDAR sensor, and an initial map of the environment. The robot will be fitted with a basic learning-based perception module that it then continuously adapts and improves while exploring the environment. Ideally, this will result in improved semantic analysis of the scene over a longer time deployment of the mobile robot or more accurate and robust robot localization.
The successful candidate will have access to ASL's high-end infrastructure including various robotic systems, GPU cluster, own office space, and the possibility to participate in field testing campaigns with the research group.
Capabilities of robotic perception systems have greatly improved since the advent of modern deep-learning based algorithms. One major drawback however, are their degraded performance under changing conditions or in new environments, preventing us from creating truly adaptive and reliable robots. In this project, we will enable our robots to continuously adapt to new environments given cameras, LiDAR sensor, and an initial map of the environment. The robot will be fitted with a basic learning-based perception module that it then continuously adapts and improves while exploring the environment. Ideally, this will result in improved semantic analysis of the scene over a longer time deployment of the mobile robot or more accurate and robust robot localization. The successful candidate will have access to ASL's high-end infrastructure including various robotic systems, GPU cluster, own office space, and the possibility to participate in field testing campaigns with the research group.
- Familiarize yourself with the current state of the art.
- Investigate existing solutions for self-supervised and continuous learning.
- Develop novel algorithms to address the challenges of online learning in real-world applications and automatic data harvesting from robotic systems.
- Deploy and test your algorithms on cutting-edge robotic systems.
- Familiarize yourself with the current state of the art. - Investigate existing solutions for self-supervised and continuous learning. - Develop novel algorithms to address the challenges of online learning in real-world applications and automatic data harvesting from robotic systems. - Deploy and test your algorithms on cutting-edge robotic systems.
- Good understanding of algorithmic challenges.
- Knowledge of Python is mandatory, C++ is recommended.
- Knowledge of ROS and a Deep Learning framework is recommended.
- Knowledge in two of the following areas: Deep Learning, machine learning, localization, sensor fusion, computer vision.
- Be curious about pushing the limits of today's intelligent robots.
- Strong self-motivation and critical mind.
- Students from outside of D-MAVT (particularly, from D-INFK, D-ITET, D-PHYS, and D-MATH) are also highly encouraged to apply.
- Good understanding of algorithmic challenges. - Knowledge of Python is mandatory, C++ is recommended. - Knowledge of ROS and a Deep Learning framework is recommended. - Knowledge in two of the following areas: Deep Learning, machine learning, localization, sensor fusion, computer vision. - Be curious about pushing the limits of today's intelligent robots. - Strong self-motivation and critical mind. - Students from outside of D-MAVT (particularly, from D-INFK, D-ITET, D-PHYS, and D-MATH) are also highly encouraged to apply.
Please send your cv and transcripts to: Abel Gawel (gawela@ethz.ch), and Hermann Blum (blumh@ethz.ch). **Please do not apply via SiROP**.
Please send your cv and transcripts to: Abel Gawel (gawela@ethz.ch), and Hermann Blum (blumh@ethz.ch). **Please do not apply via SiROP**.