Uploaded October 2021 | Updated September 2026, 2 hours ago
Do Good Robotics Symposium: Probabilistic and Machine Learning Approaches for Autonomous Robots and Automated Driving
Wolfram Burgard
VP for Automated Driving Technology
Toyota Research Institute in Los Altos, USA
Autonomous robots and and self-driving cars are currently regarded as key technologies for changing our everyday life. They are envisiond to make production and logistic processes more effective and to increase the efficiency and safety of daily traffic. In this presentation I will describe the probabilistic foundations of building such autonomous systems. In addition, I will present how modern techniques from machine learning can be used to achieve advanced perception capabilities for such systems. I will focus on two key problems in this context. First, I will describe how to deal with the high dimensionality of the underlying state estimation processes. Second, I will present how to perform learning tasks in an unsupervised fashion to resolve the burden of manually labeling huge data sets. I will present several practical applications in the context of autonomous robots and cars that drive in an automated fashion.
For more information on the Do Good Robotics Symposium see:
https://dgrs.umd.edu/
Do Good Robotics Symposium: Probabilistic and Machine Learning Approaches for Autonomous Robots and Automated Driving
Wolfram Burgard
VP for Automated Driving Technology
Toyota Research Institute in Los Altos, USA
Autonomous robots and and self-driving cars are currently regarded as key technologies for changing our everyday life. They are envisiond to make production and logistic processes more effective and to increase the efficiency and safety of daily traffic. In this presentation I will describe the probabilistic foundations of building such autonomous systems. In addition, I will present how modern techniques from machine learning can be used to achieve advanced perception capabilities for such systems. I will focus on two key problems in this context. First, I will describe how to deal with the high dimensionality of the underlying state estimation processes. Second, I will present how to perform learning tasks in an unsupervised fashion to resolve the burden of manually labeling huge data sets. I will present several practical applications in the context of autonomous robots and cars that drive in an automated fashion.
For more information on the Do Good Robotics Symposium see:
https://dgrs.umd.edu/
