This work builds upon existing task and motion planning (TAMP) frameworks by integrating pre-trained Sequencing Task-Agnostic Policies (STAP) and Effort Level Search (ELS) to create a hierarchical approach that decouples high-level task decisions from low-level motion execution. The method enhances the planning process by incorporating a novel success rate estimator (P ), which provides more accurate task success predictions than traditional Q-value estimators. We formalize the problem of long-horizon manipulation tasks, where high-level decisions are made in discrete spaces and low-level actions are executed in continuous space. To guide the search process efficiently, we leverage the infinite completion tree structure of ELS, which dynamically adjusts computational resources based on task complexity. Empirical results demonstrate that our approach significantly improves planning efficiency and execution reliability, outperforming traditional methods by reducing the search space and computational overhead. Our work highlights the effectiveness of combining learned skills from STAP with ELS and P in a hierarchical structure, laying the foundation for scalable robotic planning in complex, real-world manipulation tasks.

Meta-reinforcement learning (meta-RL) is a promising framework for tackling challenging domains requiring efficient exploration. Existing meta-RL algorithms are characterized by low sample efficiency, and mostly focus on low-dimensional task distributions. In parallel, model-based RL methods have been successful in solving partially observable MDPs, of which meta-RL is a special case. In this work, we leverage this success and propose a new model-based approach to meta-RL, based on elements from existing state-of-the-art model-based and meta-RL methods. We demonstrate the effectiveness of our approach on common meta-RL benchmark domains, attaining greater return with better sample efficiency (up to $15\times$) while requiring very little hyperparameter tuning. In addition, we validate our approach on a slate of more challenging, higher-dimensional domains, taking a step towards real-world generalizing agents.

Many AI problems, in robotics and other domains, are goal-directed, essentially seeking a trajectory leading to some goal state. Reinforcement learning (RL), building on Bellman’s optimality equation, naturally optimizes for a single goal, yet can be made goal-directed by augmenting the state with the goal. Instead, we propose a new RL framework, derived from a dynamic programming equation for the all pairs shortest path (APSP) problem, which naturally solves goal-directed queries. We show that this approach has computational benefits for both standard and approximate dynamic programming. Interestingly, our formulation prescribes a novel protocol for computing a trajectory: instead of predicting the next state given its predecessor, as in standard RL, a goal-conditioned trajectory is constructed by first predicting an intermediate state between start and goal, partitioning the trajectory into two. Then, recursively, predicting intermediate points on each sub-segment, until a complete trajectory is obtained. We call this trajectory structure a sub-goal tree. Building on it, we additionally extend the policy gradient methodology to recursively predict sub-goals, resulting in novel goal-based algorithms. Finally, we apply our method to neural motion planning, where we demonstrate significant improvements compared to standard RL on navigating a 7-DoF robot arm between obstacles.

In this work we propose a top-down decision tree learning algorithm with a class of linear classifiers called stochastic linear classifiers as the internal nodes’ hypothesis class. To this end, we derive efficient algorithms for minimizing the Gini index for this class for each internal node, although the problem is non-convex. Moreover, the proposed algorithm has a theoretical guarantee under the weak stochastic hypothesis assumption.