Online help systems: Learning while working

Most widely-used computer software packages, such as word processors, spreadsheets and web browsers, incorporate comprehensive help systems, partly because the software is meant for those with little technical knowledge. This paper identifies four systematic philosophies or approaches to help system delivery, namely the documentation approach, based on written documents, either paper-based or online; the training approach, either offered before the user starts working on the software or on-the-job; intelligent help, that is online, context-sensitive help or that relying on software agents; and finally an approach based on minimalism, defined as providing help only when and where it is needed.

Improved joint control using a genetic algorithm for a humanoid robot

This paper describes experiments conducted in order to simultaneously tune 15 joints of a humanoid robot. Two Genetic Algorithm (GA) based tuning methods were developed and compared against a hand-tuned solution. The system was tuned in order to minimise tracking error while at the same time achieve smooth joint motion. Joint smoothness is crucial for the accurate calculation of online ZMP estimation, a prerequisite for a closedloop dynamically stable humanoid walking gait. Results in both simulation and on a real robot are presented, demonstrating the superior smoothness performance of the GA based methods.

A helicopter named Dolly: Behavioural cloning for autonomous helicopter control

This paper considers the pros and cons of using behavioural cloning for the development of low-level helicopter automation modules. Over the course of this project several Behavioural cloning approaches have been investigated. The results of the most effective Behavioural cloning approach are then compared to PID modules designed for the same aircraft. The comparison takes into consideration development time, reliability, and control performance. It has been found that Behavioural cloning techniques employing local approximators and a wide state-space coverage during training can produce stabilising control modules in less time than tuning PID controllers. However, performance and reliabity deficits have been found to exist with the Behavioural Cloning, attributable largely to the time variant nature of the dynamics due to the operating environment, and the pilot actions being poor for teaching. The final conclusion drawn here is that tuning PID modules remains superior to behavioural cloning for low-level helicopter automation.

Helicopter automation using a low-cost sensing system

This paper details the design of an autonomous helicopter control system using a low cost sensor suite. Control is maintained using simple nested PID loops. Aircraft attitude, velocity, and height is estimated using an in-house designed IMU and vision system. Information is combined using complimentary filtering. The aircraft is shown to be stabilised and responding to high level demands on all axes, including heading, height, lateral velocity and longitudinal velocity.

Knowledge maintenance and the frame problem

Knowledge maintenance is a major challenge for both knowledge management and the Semantic Web. Operating over the Semantic Web, there will be a network of collaborating agents, each with their own ontologies or knowledge bases. Change in the knowledge state of one agent may need to be propagated across a number of agents and their associated ontologies. The challenge is to decide how to propagate a change of knowledge state. The effects of a change in knowledge state cannot be known in advance, and so an agent cannot know who should be informed unless it adopts a simple ‘tell everyone – everything’ strategy. This situation is highly reminiscent of the classic Frame Problem in AI. We argue that for agent-based technologies to succeed, far greater attention must be given to creating an appropriate model for knowledge update. In a closed system, simple strategies are possible (e.g. ‘sleeping dog’ or ‘cheap test’ or even complete checking). However, in an open system where cause and effect are unpredictable, a coherent cost-benefit based model of agent interaction is essential. Otherwise, the effectiveness of every act of knowledge update/maintenance is brought into question.

Sustained release of the CCR5 inhibitors CMPD167 and maraviroc from vaginal rings in rhesus macaques

Antiretroviral entry inhibitors are now being considered as vaginally administered microbicide candidates for the prevention of the sexual transmission of human immunodeficiency virus. Previous studies testing the entry inhibitors maraviroc and CMPD167 in aqueous gel formulations showed efficacy in the macaque challenge model, although protection was highly dependent on the time period between initial gel application and subsequent challenge. In this paper, we describe the sustained release of maraviroc and CMPD167 from matrix-type silicone elastomer vaginal rings both in vitro and in vivo. Both inhibitors were released continuously during 28 days from rings in vitro at rates of 100 to 2,500 µg/day. In 28-day pharmacokinetic studies in rhesus macaques, the compounds were measured in the vaginal fluid and vaginal tissue; steady-state fluid concentrations were ~10(6)-fold greater than the 50% inhibitory concentrations (IC(50)s) for simian human immunodeficiency virus 162P3 inhibition in macaque lymphocytes in vitro. Plasma concentrations for both compounds were very low. The pretreatment of macaques with Depo-Provera (DP), which is commonly used in macaque challenge studies, was shown to significantly modify the biodistribution of the inhibitors but not the overall amount released. Vaginal fluid and tissue concentrations were significantly decreased while plasma levels increased with DP pretreatment. These observations have implications for designing macaque challenge experiments and also for ring performance during the human female menstrual cycle.