PROWLER.io’s principled decision-making platform makes it possible to perceive and affect the ways agents – such as vehicles, drones, robots, characters in games and even people – interact in complex environments. Its probabilistic models enable customers to understand, guide and optimise the millions of micro-decisions that can occur in dynamic systems. The platform also provides safe, effective ways to test and validate emerging technologies.
PROWLER.io has the potential to transform complex systems design and implementation. The platform is “principled” in the sense that its core technology is founded on mathematical principles from three previously segregated fields: probabilistic modelling, machine learning and game theory. The resulting synergy of ideas and methodologies makes possible for the first time a platform whose decisions are clearly based on interpretable principles. It accomplishes this by using: powerful statistical tools to generate flexible, dynamic probabilistic models which provide new insights about virtual or physical environments; machine learning and decision-making methodologies that are more visible and interpretable than those that take place within deep neural nets; multi-agent systems that are much more flexible, adaptable and strategically interactive than traditional decision-tree based systems. Initial applications of the platform: In game development
PROWLER.io agents supersede the use of hand-crafted rules for decision-making, which are time-consuming, expensive and restrictive. The result is games that feel truly open and responsive and engage players in novel, freer, more personalised ways. Moreover, development costs and time to market decrease when testing is handled by teams of humans working with agents that can perform dull, repetitive tasks thousands of times faster than manual testers. It is impossible to program autonomous vehicles for every eventuality they will face on the roads. The platform uses probabilistic modelling to enable a self-driving car to “understand” itself and its environment, multiple principled learning approaches to teach it to drive and multi-agent systems to ensure that it operates safely alongside other road users. In smart cities, the platform optimises fleet planning and management. This ensures that real time demand for AVs matches supply, vehicles are close by when needed, routes are planned efficiently, congestion is reduced and negative environmental impacts are minimised.