Robotic applications spread to a variety of application domains, from autonomous cars and drones to domestic robots and personal devices. Each application domain comes with a rich set of requirements such as legal policies, safety and security standards, company values, or simply public perception. They must be realised as verifiable properties of software and hardware. Consider the following policy: a self-driving car must never break the highway code.
A large variety of robotic applications strongly involve handling various objects as the core process for task completion. To date, most of these jobs are still performed by people. Although some are automated by robots, those solutions primarily rely on pre-designed rules or tele-operation (limited operational time due to cognitive overload), which unavoidably limits the performance in changing environments. This project consists of multiple challenging research topics in robotic manipulation.
Pulmonary infiltrates such as pus, blood, or protein, which lingers within the parenchyma of the lungs are the leading cause of pneumonia, tuberculosis. Pulmonary infiltrates in mechanically ventilated (MV) critically ill patients in the intensive care unit (ICU) are a major diagnostic challenge and due to the poor sampling methods available.
There is a critical need for robust and accurate tools to scale up biodiversity monitoring and to manage the impact of anthropogenic change. For example, the monitoring of individual species that are particularly sensitive to habitat conversion and climate change can act as an important indicator of ecosystem health. Existing approaches for collecting data on individual species in the wild have traditionally been based on manual surveys performed by human experts.
The shape and 3D structure of the world provides us with rich signal that enables us to interact with objects and to navigate in novel and dynamic environments. Despite the importance of this information to human visual reasoning it still remains largely underutilized in modern deep learning based semantic image understanding pipelines commonly used in robotics. For example, current best performing approaches for object classification in images are predominantly based on heavily supervised feedforward convolutional neural networks.
The coming decades will see the creation of fully autonomous vehicles (AVs) capable of driving without human intervention. Among the expected benefits of AVs are a significant reduction in traffic incidents, congestion, and pollution, while dramatically improving cost-efficiency.
Deep Neural Network (DNN) technologies coupled with GPU type hardware provide practical methods for learning complex functions from vast datasets. However, their architectures are often developed using trial and error approaches and the resulting systems normally provide ‘black box’ solutions containing many millions of learnt but abstract parameters. They are therefore extremely difficult to interpret and understand, and their accuracy and certainty of prediction, or classification, are normally not known.
Data-driven machine learning techniques are popularly used in the field of robotics to inform autonomous decision-making and to perform control or path-planning. Supervised learning and reinforcement learning have been shown to be particularly amenable to canonical tasks that are integral to robotics applications. However, these techniques rely on data in the form of action-label (supervised), action-value (regression) or action-reward (RL) pairs, where the action is a path (or some other) execution by a real robot. e.g.
Curiosity guides humans to learn efficiently. It incentivizes us to spend more energy and time examining new, unexpected things, and to disregard those we fully understand already, to make our learning more efficient. Much of the vision learning that is done today is passive: learning systems are exposed to large amounts of training data, and learn from each sample multiple times, regardless of their current ability to recognize them at the time. This makes the process slow, specially given the increasingly large number of samples on datasets.