How can autonomous robots intelligently perceive, understand and interact with their environment?
To answer this long-term research question, we focus on the following four areas of robotics research.
How to fuse measurements obtained from the robot's on-board sensors, and those obtained from communication with other robots in the team to estimate the state of the environment? Sensors primarily include the onboard cameras, as well as other sensors such as IMUs, event cameras, air-speed sensor, etc. State of the environment includes positions and poses of subjects such as humans, animals and other robots, as well as the map of the static part of the environment. Click here to read more.
How to infer meaning from the estimated state, or directly from the raw sensor measurements, and make high-level decisions based on that? Examples include inferring behavior of the tracked
subjects (animals or humans), predict the future motion and trajectory of the subjects, infer risk of collision with the subjects or the static environment, etc.
Furthermore, based on the perceived state of the environment or/and inferred situation, how to control the robots to accomplish various given tasks? Tasks could include, for example, monitoring/following a group of subjects, or mapping a desired location or transporting objects between locations. Click here to read more.
How can autonomous robots perceive, understand and interact with their environment, given i) very little prior information about the environment, ii) little knowledge about its own system dynamics and iii) dynamic environments with changing conditions, e.g., changing visibility, occupancy, etc? Secondly, under these conditions how can these robots optimize energy consumption, computational requirements and the communication bandwidth? Can these questions be answered through learning-based methods -- specifically reinforcement learning, and learning in simulation and from synthetic data? Click here to read more.
Which novel and innovative autonomous robotic systems are needed to solve the given challenge? While this is typically an application-specific question, it has a direct impact on the development of various other methodologies. For example, differences in the dynamics of an airship and multi-rotor robot significantly impact the control methods they use. Thus, a long term effort in this context will be on the development of learning-based methods that are agnostic, or require little previous knowledge of the system. Click here to read more.