Mobile GUI agents are becoming critical tools for enhancing human-device interaction efficiency, with multimodal large language models (MLLMs) emerging as dominant paradigms in this domain. Current agents, however, are limited to following explicit human instructions, resulting in insufficient capability for proactive intent anticipation. Additionally, these agents fail to leverage the contextual information associated with users during task execution, thereby neglecting potentially vast differences in user preferences. To address these challenges, we introduce the FingerTip benchmark. It contains two new tracks: proactive task suggestions by analyzing environment observation and users' previous intents, and personalized task execution by catering to users' action preferences. We collected unique human demonstrations of multi-step Android device interactions across a variety of everyday apps. These demonstrations are not isolated but are continuously acquired from the users' long-term usage in their real lives, and encompass essential user-related contextual information. Our experiments reveal challenges of the tasks we propose. The model fine-tuned with the data we collected effectively utilized user information and achieved good results, highlighting the potential of our approach in building more user-oriented mobile GUI agents. Our code is open-source at https://anonymous.4open.science/r/FingerTip-57B8 for reproducibility.

To solve the problem of lateral and logitudinal joint decision-making of multi-vehicle cooperative driving for connected and automated vehicles (CAVs), this paper proposes a Monte Carlo tree search (MCTS) method with parallel update for multi-agent Markov game with limited horizon and time discounted setting. By analyzing the parallel actions in the multi-vehicle joint action space in the partial-steady-state traffic flow, the parallel update method can quickly exclude potential dangerous actions, thereby increasing the search depth without sacrificing the search breadth. The proposed method is tested in a large number of randomly generated traffic flow. The experiment results show that the algorithm has good robustness and better performance than the SOTA reinforcement learning algorithms and heuristic methods. The vehicle driving strategy using the proposed algorithm shows rationality beyond human drivers, and has advantages in traffic efficiency and safety in the coordinating zone.

Multi-robot belief space planning (MR-BSP) is essential for reliable and safe autonomy. While planning, each robot maintains a belief over the state of the environment and reasons how the belief would evolve in the future for different candidate actions. Yet, existing MR-BSP works have a common assumption that the beliefs of different robots are consistent at planning time. Such an assumption is often highly unrealistic, as it requires prohibitively extensive and frequent communication capabilities. In practice, each robot may have a different belief about the state of the environment. Crucially, when the beliefs of different robots are inconsistent, state-of-the-art MR-BSP approaches could result in a lack of coordination between the robots, and in general, could yield dangerous, unsafe and sub-optimal decisions. In this paper, we tackle this crucial gap. We develop a novel decentralized algorithm that is guaranteed to find a consistent joint action. For a given robot, our algorithm reasons for action preferences about 1) its local information, 2) what it perceives about the reasoning of the other robot, and 3) what it perceives about the reasoning of itself perceived by the other robot. This algorithm finds a consistent joint action whenever these steps yield the same best joint action obtained by reasoning about action preferences; otherwise, it self-triggers communication between the robots. Experimental results show efficacy of our algorithm in comparison with two baseline algorithms.

Training practical agents usually involve offline and online reinforcement learning (RL) to balance the policy's performance and interaction costs. In particular, online fine-tuning has become a commonly used method to correct the erroneous estimates of out-of-distribution data learned in the offline training phase. However, even limited online interactions can be inaccessible or catastrophic for high-stake scenarios like healthcare and autonomous driving. In this work, we introduce an interaction-free training scheme dubbed Offline-with-Action-Preferences (OAP). The main insight is that, compared to online fine-tuning, querying the preferences between pre-collected and learned actions can be equally or even more helpful to the erroneous estimate problem. By adaptively encouraging or suppressing policy constraint according to action preferences, OAP could distinguish overestimation from beneficial policy improvement and thus attains a more accurate evaluation of unseen data. Theoretically, we prove a lower bound of the behavior policy's performance improvement brought by OAP. Moreover, comprehensive experiments on the D4RL benchmark and state-of-the-art algorithms demonstrate that OAP yields higher (29% on average) scores, especially on challenging AntMaze tasks (98% higher).

Policy gradient (PG) estimators for softmax policies are ineffective with sub-optimally saturated initialization, which happens when the density concentrates on a sub-optimal action. Sub-optimal policy saturation may arise from bad policy initialization or sudden changes in the environment that occur after the policy has already converged, and softmax PG estimators require a large number of updates to recover an effective policy. This severe issue causes high sample inefficiency and poor adaptability to new situations. To mitigate this problem, we propose a novel policy gradient estimator for softmax policies that utilizes the bias in the critic estimate and the noise present in the reward signal to escape the saturated regions of the policy parameter space. Our analysis and experiments, conducted on bandits and classical MDP benchmarking tasks, show that our estimator is more robust to policy saturation.

We develop a normative framework for hierarchical model-based policy optimization based on applying second-order methods in the space of all possible state-action paths. The resulting natural path gradient performs policy updates in a manner which is sensitive to the long-range correlational structure of the induced stationary state-action densities. We demonstrate that the natural path gradient can be computed exactly given an environment dynamics model and depends on expressions akin to higher-order successor representations. In simulation, we show that the priorization of local policy updates in the resulting policy flow indeed reflects the intuitive state-space hierarchy in several toy problems.

We advocate a case-supported principle-based behavior paradigm coupled with the Fractal robot architecture as a means to control an eldercare robot. The most ethically preferable action at any given moment is determined using a principle, abstracted from cases where a consensus of ethicists exists.