Predicate Invention for Bilevel Planning

Silver, Tom; Chitnis, Rohan; Kumar, Nitin; McClinton, Willie; Lozano-Pérez, Tomás; Kaelbling, Leslie Pack; Tenenbaum, Joshua B.

doi:10.1609/aaai.v37i10.26429

Cited by 4 publications

(2 citation statements)

References 41 publications

(72 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also tried the Slitherlink domain, which was featured in the 2023 International Planning Competition, and found that GPT-4 did not recognize that no simple strategy exists (Takayuki 2000). Generalized planning without LLMs also remains important in cases where domain descriptions are not human-readable, e.g., because the predicates or operators are learned (Silver et al 2023). Even with natural language descriptions, combining "classical" approaches with LLMs may be best.…”

Section: Discussion and Future Workmentioning

confidence: 99%

Generalized Planning in PDDL Domains with Pretrained Large Language Models

Silver,

Dan,

Srinivas

et al. 2024

AAAI

View full text Add to dashboard Cite

Recent work has considered whether large language models (LLMs) can function as planners: given a task, generate a plan. We investigate whether LLMs can serve as generalized planners: given a domain and training tasks, generate a program that efficiently produces plans for other tasks in the domain. In particular, we consider PDDL domains and use GPT-4 to synthesize Python programs. We also consider (1) Chain-of-Thought (CoT) summarization, where the LLM is prompted to summarize the domain and propose a strategy in words before synthesizing the program; and (2) automated debugging, where the program is validated with respect to the training tasks, and in case of errors, the LLM is re-prompted with four types of feedback. We evaluate this approach in seven PDDL domains and compare it to four ablations and four baselines. Overall, we find that GPT-4 is a surprisingly powerful generalized planner. We also conclude that automated debugging is very important, that CoT summarization has non-uniform impact, that GPT-4 is far superior to GPT-3.5, and that just two training tasks are often sufficient for strong generalization.

show abstract

Section: Discussion and Future Workmentioning

confidence: 99%

Generalized Planning in PDDL Domains with Pretrained Large Language Models

Silver,

Dan,

Srinivas

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…In another line work, [20], [21], [22] proposed a bilevel planning schema in which a set of operators and corresponding samplers are learned from previously acquired symbols that allow the agent to make refined plans that can consider the geometric information. [23] learns predicates from demonstrations with a surrogate objective for planning. [24] learns policies parameterized by neural networks based on the symbolic state transitions.…”

Section: Related Workmentioning

confidence: 99%

DeepSym: Deep Symbol Generation and Rule Learning for Planning from Unsupervised Robot Interaction

Ahmetoğlu

Seker

Piater

et al. 2022

jair

View full text Add to dashboard Cite

Symbolic planning and reasoning are powerful tools for robots tackling complex tasks. However, the need to manually design the symbols restrict their applicability, especially for robots that are expected to act in open-ended environments. Therefore symbol formation and rule extraction should be considered part of robot learning, which, when done properly, will offer scalability, flexibility, and robustness. Towards this goal, we propose a novel general method that finds action-grounded, discrete object and effect categories and builds probabilistic rules over them for non-trivial action planning. Our robot interacts with objects using an initial action repertoire that is assumed to be acquired earlier and observes the effects it can create in the environment. To form action-grounded object, effect, and relational categories, we employ a binary bottleneck layer in a predictive, deep encoderdecoder network that takes the image of the scene and the action applied as input, and generates the resulting effects in the scene in pixel coordinates. After learning, the binary latent vector represents action-driven object categories based on the interaction experience of the robot. To distill the knowledge represented by the neural network into rules useful for symbolic reasoning, a decision tree is trained to reproduce its decoder function. Probabilistic rules are extracted from the decision paths of the tree and are represented in the Probabilistic Planning Domain Definition Language (PPDDL), allowing off-the-shelf planners to operate on the knowledge extracted from the sensorimotor experience of the robot. The deployment of the proposed approach for a simulated robotic manipulator enabled the discovery of discrete representations of object properties such as ‘rollable’ and ‘insertable’. In turn, the use of these representations as symbols allowed the generation of effective plans for achieving goals, such as building towers of the desired height, demonstrating the effectiveness of the approach for multi-step object manipulation. Finally, we demonstrate that the system is not only restricted to the robotics domain by assessing its applicability to the MNIST 8-puzzle domain in which learned symbols allow for the generation of plans that move the empty tile into any given position.

show abstract

LEAGUE: Guided Skill Learning and Abstraction for Long-Horizon Manipulation

Cheng,

2023

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

Predicate Invention for Bilevel Planning

Cited by 4 publications

References 41 publications

Generalized Planning in PDDL Domains with Pretrained Large Language Models

Generalized Planning in PDDL Domains with Pretrained Large Language Models

DeepSym: Deep Symbol Generation and Rule Learning for Planning from Unsupervised Robot Interaction

LEAGUE: Guided Skill Learning and Abstraction for Long-Horizon Manipulation

Contact Info

Product

Resources

About