How can we plan efficiently in a large and complex environment when the time budget is limited? Given the original simulator of the environment, which may be computationally very demanding, we propose to learn online an approximate but much faster simulator that improves over time. To plan reliably and efficiently while the approximate simulator is learning, we develop a method that adaptively decides which simulator to use for every simulation, based on a statistic that measures the accuracy of the approximate simulator. This allows us to use the approximate simulator to replace the original simulator for faster simulations when it is accurate enough under the current context, thus trading off simulation speed and accuracy. Experimental results in two large domains show that when integrated with POMCP, our approach allows to plan with improving efficiency over time.@en

The imperfect decision-making of human buyers participating in retail markets varies from fundamental models that assume rational economic choices: even in markets with identical items human buyers are not rational, i.e., buyers do not always choose the cheapest option. Recent developments in artificial intelligence and e-commerce enable market participation by software agents that are (almost) perfectly rational due to their computational capacity. However, the increasing degree of buyers’ rationality might have unfavorable effects on retail markets with regards to the competition between sellers and the resulting prices. In this paper, we study the effects of varying degrees of buyers’ rationality on the competition and the prices buyers face in retail markets with identical items. We use the multinomial logit function to model different degrees of buyers’ rationality. We further model the competition between sellers using k-level reasoning: each seller computes the price to offer (best response strategy) with regards to its belief for the competition. First, we derive an analytical best response strategy (price) of a seller given the competing prices and the degree of buyers’ rationality, and show that there exists an optimal degree of buyers’ rationality that minimizes the price. Last, we use evolutionary game theory to show that perfect rationality leads to unstable competition dynamics increasing the overall cost for buyers. In contrast, bounded rationality leads to smoother dynamics and lower cost for buyers. Our insights raise the need to revisit design objectives for software agents in retail markets in light of their wider systematic impact.

@en

We study mechanisms to incentivize demand response in smart energy systems. We assume agents that can respond (reduce their demand) with some probability if they prepare prior to the real-ization of the demand. Both preparation and response incur costs to agents. Previous work studies truthful mechanisms that select a minimal set of agents to prepare and respond such that a fixed demand reduction target is achieved with high probability. In this work we additionally consider the balancing responsibility of a retailer under a given demand forecast and imbalance price: The retailer is responsible to purchase additional reserve capacity at a high imbalance price to cover any excess in the demand. In this extended setting we study mechanisms that request only a subset of prepared agents to respond since the reduction target depends on the realization of the demand: We propose: (i) a sequential mechanism that in each round embeds a second-price auction and is truthful under some mild assumptions for the setting, and (ii) a truthful combinatorial mechanism that runs in polynomial time and uses VCG payments. We show that both mechanisms guarantee non-negative utility in expectation for both agents and the retailer (mechanism), and can further be used for simultaneous downward and upward flexibility. Last, we verify our theoretical findings in an empirical evaluation over a wide range of mechanism parameters.

@en

Computers that negotiate on our behalf hold great promise for the future and will even become indispensable in emerging application domains such as the smart grid and the Internet of Things. Much research has thus been expended to create agents that are able to negotiate in an abundance of circumstances. However, up until now, truly autonomous negotiators have rarely been deployed in real-world applications. This paper sizes up current negotiating agents and explores a number of technological, societal and ethical challenges that autonomous negotiation systems have brought about. The questions we address are: in what sense are these systems autonomous, what has been holding back their further proliferation, and is their spread something we should encourage? We relate the automated negotiation research agenda to dimensions of autonomy and distill three major themes that we believe will propel autonomous negotiation forward: accurate representation, long-term perspective, and user trust. We argue these orthogonal research directions need to be aligned and advanced in unison to sustain tangible progress in the field. @en

Computers that negotiate on our behalf hold great promise for the future and will even become indispensable in emerging application domains such as the smart grid, autonomous driving, and the Internet of Things. Much research has thus been expended to create agents that are able to negotiate in an abundance of circumstances. However, up until now, truly autonomous negotiators have rarely been deployed in real-world applications. This paper sizes up current negotiating agents and explores a number of technological, societal and ethical challenges that autonomous negotiation systems are bringing about. The questions we address are: in what sense are these systems autonomous, what has been holding back their further proliferation, and is their spread something we should encourage? We relate the automated negotiation research agenda to dimensions of autonomy and distill three major themes that we believe will propel autonomous negotiation forward: accurate representation, long-term perspective, and user trust. We argue these orthogonal research directions need to be aligned and advanced in unison to sustain tangible progress in the field.@en

We present a unified model for flexibility services in the power system, identify two existing categories (ramping and loading) and introduce a new category (stalling). Each service is characterised by duration, capacity and effort, with associated prices. We show that the effort of stalling, measurable in kWh2, is a significant cost component for balancing through storage and demand response (DR). In future energy systems - with increased reliance on renewable generation - storage and DR resources are expected to become an important component of power system flexibility and balancing. Existing resource allocation mechanisms are mainly based on pricing in kW (for ramping) and in kWh (for load). Simulations demonstrate that conventional pricing mechanisms yield inefficient allocations of storage and DR resources for power balancing. In contrast, we show that introducing an additional pricing component in e/kWh2 (for stalling) improves the allocation efficiency.@en

Current electricity tariffs for retail rarely provideincentives for intelligent demand response of flexiblecustomers. Such customers could otherwisecontribute to balancing supply and demand in futuresmart grids. This paper proposes an innovativerisk-sharing tariff to incentivize intelligentcustomer behavior. A two-step parameterized paymentscheme is proposed, consisting of a prepaymentbased on the expected consumption, and asupplementary payment for any observed deviationfrom the anticipated consumption. Within a gametheoreticalanalysis, we capture the strategic con-flict of interest between a retailer and a customerin a two-player game, and we present optimal, i.e.,best response, strategies for both players in thisgame. We show analytically that the proposed tariffprovides customers of varying flexibility with variableincentives to assume and alleviate a fraction ofthe balancing risk, contributing in this way to theuncertainty reduction in the envisioned smart-grid.@en

Smart energy systems integrate renewables and demand response. Most European electricity markets coordinate the resulting time-varying flexibility in demand and supply by organising day-ahead trade with Walrasian mechanisms, using simultaneous call auctions and sealed bids. These mechanisms give bidders no information on each other's values and flexibilities until after clearing. In this paper we simulate two alternative day-ahead market mechanisms which share information, such that bidders obtain a better position before entering the intraday market. One mechanism uses an ascending shared market price signal rather than sealed bids. The other auctions off future timeslots consecutively rather than simultaneously. We perform a case study on 400 households with electric vehicles, either with or without volatile wind generation. Results show that a price-taking flexible consumer can obtain higher utility in the market with simultaneous ascending-price auctions, because online price information reduces uncertainty over available energy and prices.@en