Peer reviewed conference papers.

Odour localisation in an enclosed area is difficult due to the formation of sectors of circulating airflow. Well-defined plumes do not exist, and reactive plume following may not be possible. Odour localisation has been partially achieved in this environment by using knowledge of airflow, and a search that relies on chemical sensing and reasoning. However the results are not specific, with the odour source only restricted to a broad area. This paper presents a solution to the problem by introducing a second search stage using visual sensing. It therefore comprises a bi-modal, two-stage search, with each stage exploiting complementary sensing modalities. This paper presents details of the method and experimental results.

Previous work on odour localisation in enclosed environments, relying on an airflow model, has faced significant limitations due to the fact that large differences between airflow topologies are predicted for only small variations in a physical map. This is due to uncertainties in the map and approximations in the modelling process. Furthermore, there are uncertainties regarding the flow direction through inlet/outlet ducts. We have presented a method for dealing with these uncertainties, by generating multiple airflow hypotheses. As the robot performs odour localisation, airflow in the environment is measured and used to adjust the confidences of the hypotheses using Bayesian inference. The best hypothesis is then selected, which allows the completion of the localisation task. We have shown experimentally that this method is capable of improving the robustness of our method for odour localisation in the presence of uncertainties, where previously it was incapable. The results further demonstrate the usefulness of Naive Physics for practical robotics applications.

Controlling finger impedance is critical for successful grasping. Understanding how humans achieve this is of great interest for learning about human motor control, as well as for applications in robotic grasping. There have been a number of studies on finger impedance in both the robotics and biological fields. They almost exclusively consider only stiffness and viscosity. However, inertia may play an important role in certain grasps, and is important for calculation of the other impedance properties. This paper reports current progress of a project to create a geometric model of the hand for predicting hand/grasp inertia at different configurations (sensed by a glove that measures joint angles) during a variety of tasks.

To date, robotics has had limited success at operating in unstructured environments. Part of the problem is the lack of commonsense reasoning. One area of commonsense reasoning is Naive Physics, the practice of using intuitive rules to reason about the physical environment. Researchers have explored relevant philosophical issues, attempted to develop logical formalisations, and developed systems to `understand' and learn simple physical/spatial qualities of physical objects. This paper reports the implementation of an algorithm that uses Naive Physics to model air°ow, and has been used on a practical robot for the task of odour localisation. This is an example of a Naive Reasoning Machine, an algorithm encapsulating naive rules, which represents a broader and more practical approach to naive physics.

This paper presents a new intelligent odor localization strategy, which enables a robot to locate the source of an odor in a cluttered indoor environment. Traditionally, work in this area has focused on open areas free of obstacles and having no walls or possessing walls without openings. Existing solutions predominantly use reactive algorithms to navigate along the entire length of the odor plume to the source. Not only is this slow, but in a cluttered indoor environment it may not be possible. In a constrained environment, airflows tend to circulate in sectors and well-defined plumes that lead upwind to the odor source do not exist. We have developed a sense-map-plan-act style control strategy to model the airflow in the environment using naï ve physics, then use the model to reason about odor dispersal, move to key positions gathering information, and make a prediction of the most likely location for an odor source. The control strategy has located the odor source for a variety of room configurations. This paper describes details of the control strategy, practical experiments, and results.

Keywords: electronic noses inference mechanisms mobile robots path planning airflows cluttered indoor environment constrained environment intelligent odor localization strategy naï ve physics navigation odor dispersal odor plume odor source location reasoning robot room configurations sense-map-plan-act style control strategy

This paper describes current progress of a project that uses naïve physics to enable a robot to perform efficient odour localisation. Odour localisation is the problem of finding the source of an odour or other volatile chemical. Performing this effectively could lead to many humanitarian and other valuable applications. Current techniques utilise reactive control schemes requiring the robot to follow the plume along its entire length, which is slow and may be especially difficult in a cluttered environment. This research is concerned with creating a more ‘intelligent’ system to overcome these limitations. A map of the robot’s environment was used, together with a naïve physics model of airflow to predict the pattern of air movement. The robot used the airflow pattern to reason about the probable location of the odour source. A prototype system was successful in a simplified cluttered environment, locating the source comparatively quickly. This demonstrates that naïve physics can be used for effective odour localisation, and has the potential to allow a robots operating in unstructured environments to reason about their surroundings. This paper presents details of the naïve physical model of airflow, reasoning system, experimental work, and results of practical odour source localisation experiments.