While mathematically elegant, a fundamental issue with AIXI is that it is only asymptotically computable.

Imitation learning aims to extract high-performance policies from logged demonstrations of expert behavior.

The advancement of embodied intelligence is accelerating the integration of robots into daily life as human assistants. This evolution requires robots to not only interpret high-level instructions and plan tasks but also perceive and adapt within dynamic environments. Vision-Language Models (VLMs) present a promising solution by combining visual understanding and language reasoning. However, existing VLM-based methods struggle with interactive exploration, accurate perception, and real-time plan adaptation. To address these challenges, we propose ExploreVLM, a novel closed-loop task planning framework powered by Vision-Language Models (VLMs). The framework is built around a step-wise feedback mechanism that enables real-time plan adjustment and supports interactive exploration. At its core is a dual-stage task planner with self-reflection, enhanced by an object-centric spatial relation graph that provides structured, language-grounded scene representations to guide perception and planning. An execution validator supports the closed loop by verifying each action and triggering re-planning. Extensive real-world experiments demonstrate that ExploreVLM significantly outperforms state-of-the-art baselines, particularly in exploration-centric tasks. Ablation studies further validate the critical role of the reflective planner and structured perception in achieving robust and efficient task execution.

Learning-based methods have shown promising performance for accelerating motion planning, but mostly in the setting of static environments.

We propose new width-based planning and learning algorithms inspired from a careful analysis of the design decisions made by previous width-based planners.

Addressing large-scale planning problems has become one of the central challenges in the planning community, deriving from the state-space explosion caused by growing objects and actions. Recently, researchers have explored the effectiveness of leveraging Large Language Models (LLMs) to generate helpful actions and states to prune the search space. However, prior works have largely overlooked integrating LLMs with domain-specific knowledge to ensure valid plans. In this paper, we propose a novel LLM-assisted planner integrated with problem decomposition, which first decomposes large planning problems into multiple simpler sub-tasks. Then we explore two novel paradigms to utilize LLMs, i.e., LLM4Inspire and LLM4Predict, to assist problem decomposition, where LLM4Inspire provides heuristic guidance according to general knowledge and LLM4Predict employs domain-specific knowledge to infer intermediate conditions. We empirically validate the effectiveness of our planner across multiple domains, demonstrating the ability of search space partition when solving large-scale planning problems. The experimental results show that LLMs effectively locate feasible solutions when pruning the search space, where infusing domain-specific knowledge into LLMs, i.e., LLM4Predict, holds particular promise compared with LLM4Inspire, which offers general knowledge within LLMs.

Landmarks$\unicode{x2013}$conditions that must be satisfied at some point in every solution plan$\unicode{x2013}$have contributed to major advancements in classical planning, but they have seldom been used in stochastic domains. We formalize probabilistic landmarks and adapt the UCT algorithm to leverage them as subgoals to decompose MDPs; core to the adaptation is balancing between greedy landmark achievement and final goal achievement. Our results in benchmark domains show that well-chosen landmarks can significantly improve the performance of UCT in online probabilistic planning, while the best balance of greedy versus long-term goal achievement is problem-dependent. The results suggest that landmarks can provide helpful guidance for anytime algorithms solving MDPs.

We propose a novel concept of operations using optimal planning methods and machine learning (ML) to collect spaceborne data that is unprecedented for monitoring wildfires, process it to create new or enhanced products in the context of wildfire danger or spread monitoring, and assimilate them to improve existing, wildfire decision support tools delivered to firefighters within latency appropriate for time-critical applications. The concept is studied with respect to NASA's CYGNSS Mission, a constellation of passive microwave receivers that measure specular GNSS-R reflections despite clouds and smoke. Our planner uses a Mixed Integer Program formulation to schedule joint observation data collection and downlink for all satellites. Optimal solutions are found quickly that collect 98-100% of available observation opportunities. ML-based fire predictions that drive the planner objective are greater than 40% more correlated with ground truth than existing state-of-art. The presented case study on the TX Smokehouse Creek fire in 2024 and LA fires in 2025 represents the first high-resolution data collected by CYGNSS of active fires. Creation of Burnt Area Maps (BAM) using ML on data from active fires and BAM assimilation into NASA's Weather Research and Forecasting Model using neural nets to broadcast fire spread are novel outcomes. BAM and CYGNSS obtained soil moisture are integrated for the first time into USGS fire danger maps. Inclusion of CYGNSS data in ML-based burn predictions boosts accuracy by 13%, and inclusion of high-resolution data boosts ML recall by another 15%. The proposed workflow has an expected latency of 6-30h, improving on the current delivery time of multiple days. All components in the proposed concept are shown to be computationally scalable and globally generalizable, with sustainability considerations such as edge efficiency and low latency on small devices.

The machines gain possibly different utilities by processing different jobs, and all jobs assigned to the same machine should be processed without overlap.