William Curran

Towards recognizing "cool": can end users help computer vision recognize subjective attributes of objects in images?

Recent computer vision approaches are aimed at richer image interpretations that extend the standard recognition of objects in images (e.g., cars) to also recognize object attributes (e.g., cylindrical, has-stripes, wet). However, the more idiosyncratic and abstract the notion of an object attribute (e.g., cool car), the more challenging the task of attribute recognition. This paper considers whether end users can help vision algorithms recognize highly idiosyncratic attributes, referred to here as subjective attributes. We empirically investigated how end users recognized three subjective attributes of carscool, cute, and classic. Our results suggest the feasibility of vision algorithms recognizing subjective attributes of objects, but an interactive approach beyond standard supervised learning from labeled training examples is needed.

Mini-crowdsourcing end-user assessment of intelligent assistants: A cost-benefit study

Intelligent assistants sometimes handle tasks too important to be trusted implicitly. End users can establish trust via systematic assessment, but such assessment is costly. This paper investigates whether, when, and how bringing a small crowd of end users to bear on the assessment of an intelligent assistant is useful from a cost/benefit perspective. Our results show that a mini-crowd of testers supplied many more benefits than the obvious decrease in workload, but these benefits did not scale linearly as mini-crowd size increased - there was a point of diminishing returns where the cost-benefit ratio became less attractive.

Evaluating impact in the ROS ecosystem

The ROS ecosystem is an interconnected web of packages, nodes and people with no efficient means to compare, assess or visualize them. We develop a set of tools consisting of various metrics, a data visualization web app, and an active monitoring system. With these tools, we measure the current state of the ecosystem as well as determine where the community should direct their efforts. We also encourage the community to provide input on potential applications, additional metrics, and further improvements to address the needs of the ROS ecosystem. We incentivize this input by gamifying community contributions to the infrastructure. Encouraging user-driven improvements to the ROS infrastructure through the use of a leaderboard and friendly competition will advance ROS development and community support far into the future.

Wearable computing to enable robot microinteractions

High-level control of mobile robots currently requires use of a personal computer and traditional graphical user interface in the vast majority of cases. Such interfaces are not natural and frequently poorly suited to interacting with robots when tasking the robot to perform very short interactions requiring only brief commands. This is a problem that is especially acute for persons with disabilities. New technologies, such as Google Glass provide a variety of high quality sensors and an unobtrusive display which allows users to have a powerful robot control interface with them at all times. In this paper we present a system which provides the basic elements required in order for a user to interact with a robot using Glass. We also informally evaluate Glass as an input device, and present several examples of applications that this interface enables.

Partitioning agents and shaping their evaluation functions in air traffic problems with hard constraints

Hundreds of thousands of hours of delay, costing millions of dollars annually, are reported by US airports. The task of managing delay may be modeled as a multiagent congestion problem with agents who collectively impact the system. In this domain, agents are tightly coupled, and the environment can quickly change, making it difficult for agents to assess how they impact the system. We combine the noise reduction of fitness function shaping, the robustness of cooperative coevolutionary algorithms, and agent partitioning to perform hard constraint optimization on the congestion and reduce the delay throughout the National Air Space (NAS). Our results show that an autonomous partitioning of the agents using system features leads to up to 540x speed over simple hard constraint enforcement, as well as up to a 21% improvement in performance over a greedy scheduling solution corresponding to hundreds of hours of delay saved in a single day.

Developing Learning from Demonstration Techniques for Individuals with Physical Disabilities

Learning from demonstration research often assumes that the demonstrator can quickly give feedback or demonstrations. Individuals with severe motor disabilities are often slow and prone to human errors in demonstrations while teaching. Our work develops tools to allow persons with severe motor disabilities, who stand to benefit most from assistive robots, to train these systems. To accommodate slower feedback, we will develop a movie-reel style learning from demonstration interface. To handle human error, we will use dimensionality reduction to develop new reinforcement learning techniques.

Hierarchical simulation for complex domains: air traffic flow management

A key element in the continuing growth of air traffic is the increased use of automation. The Next Generation (Next-Gen) Air Traffic System will include automated decision support systems and satellite navigation that will let pilots know the precise locations of other aircraft around them. This Next-Gen suggestion system can assist pilots in making good decisions when they have to direct the aircraft themselves. However, effective automation is critical in achieving the capacity and safety goals of the Next-Gen Air Traffic System. In this paper we show that evolutionary algorithms can be used to achieve this effective automation. However, it is not feasible to use a standard evolutionary algorithm learning approach in such a detailed simulation. Therefore, we apply a hierarchical simulation approach to an air traffic congestion problem where agents must reach a destination while avoiding separation violations. Due to the dynamic nature of this problem, agents need to learn fast. Therefore, we apply low fidelity simulation for agents learning their destination, and a high fidelity simulation employing the Next-Gen technology for learning separation assurance. The hierarchical simulation approach increases convergence rate, leads to a better performing solution, and lowers computational complexity by up to 50 times.