Subretinal injection is a delicate and complex microsurgery. The main surgical difficulties come from the surgeon's hand tremor, dexterous motion, and insufficient visual feedback. In order to begin addressing these challenges, this paper presents a robot system for subretinal insertion integrated with intraoperative optical coherence tomography (OCT). The surgical workflow using this system consists of two main parts: the first part is the manual robot control, which aims the target before approaching the retinal surface, while considering the remote center of motion (RCM) constraint. When the injection area has been located precisely, needle is inserted into retina. To ensure \revise{surgical safety}, needle insertion depth is estimated using OCT images on a continuous basis. A soft RCM control method is designed and integrated for the controller of our hybrid parallel-serial surgical robot. Safety and accuracy performance evaluation with a 15 ms control loop shows that the worst case RCM deviation error is within 1 mm. Experimental results demonstrated that the proposed system has the ability to improve surgical outcomes by surgeons overcome their physical limitations in order to enable a better dexterous motion, and furthermore enhancing their visual feedback for a better intraocular perception.

A key challenge for autonomous driving is to process a massive amount of sensor data and make safe and reliable decisions in real-time. However, autonomous vehicles often have insufficient on-board resources to provide the required computation capacity. To address this problem, this paper advocates a novel approach to offload computation-intensive autonomous driving services to road-side units and cloud for swift executions. Our approach combines an Integer Linear Programming (ILP) formulation for offline optimization of the scheduling strategy and a fast heuristics algorithm for online adaptation. We verify our technique with both synthetic task graphs and real-world deployment. Experimental results show that our approach can improve system performance effectively.

Humans can continually accumulate knowledge and develop increasingly complex behaviours and skills throughout their lives, which is a capability known as ‘lifelong learning’. Although this lifelong learning capability is considered an essential mechanism that makes up general intelligence, recent advancements in artificial intelligence predominantly excel in narrow, specialized domains and generally lack this lifelong learning capability. Here we introduce a robotic lifelong reinforcement learning framework that addresses this gap by developing a knowledge space inspired by the Bayesian non-parametric domain. In addition, we enhance the agent’s semantic understanding of tasks by integrating language embeddings into the framework. Our proposed embodied agent can consistently accumulate knowledge from a continuous stream of one-time feeding tasks. Furthermore, our agent can tackle challenging real-world long-horizon tasks by combining and reapplying its acquired knowledge from the original tasks stream. The proposed framework advances our understanding of the robotic lifelong learning process and may inspire the development of more broadly applicable intelligence.

In recent years, the subject of deep reinforcement learning (DRL) has developed very rapidly, and is now applied in various fields, such as decision making and control tasks. However, artificial agents trained with RL algorithms require great amounts of training data, unlike humans that are able to learn new skills from very few examples. The concept of meta-reinforcement learning (meta-RL) has been recently proposed to enable agents to learn similar but new skills from a small amount of experience by leveraging a set of tasks with a shared structure. Due to the task representation learning strategy with few-shot adaptation, most recent work is limited to narrow task distributions and stationary environments, where tasks do not change within episodes. In this work, we address those limitations and introduce a training strategy that is applicable to non-stationary environments, as well as a task representation based on Gaussian mixture models to model clustered task distributions. We evaluate our method on several continuous robotic control benchmarks. Compared with state-of-the-art literature that is only applicable to stationary environments with few-shot adaption, our algorithm first achieves competitive asymptotic performance and superior sample efficiency in stationary environments with zero-shot adaption. Second, our algorithm learns to perform successfully in non-stationary settings as well as a continual learning setting, while learning well-structured task representations. Last, our algorithm learns basic distinct behaviors and well-structured task representations in task distributions with multiple qualitatively distinct tasks.