We introduce a new method that extracts knowledge from a large language model (LLM) to produce object-level plans, which describe high-level changes to object state, and uses them to bootstrap task and motion planning (TAMP). Existing work uses LLMs to directly output task plans or generate goals in representations like PDDL. However, these methods fall short because they rely on the LLM to do the actual planning or output a hard-to-satisfy goal. Our approach instead extracts knowledge from an LLM in the form of plan schemas as an object-level representation called functional object-oriented networks (FOON), from which we automatically generate PDDL subgoals. Our method markedly outperforms alternative planning strategies in completing several pick-and-place tasks in simulation.††Project Website: https://davidpaulius.github.io/olpllm/

State uncertainty is a primary obstacle to effective long-horizon robot task planning. State uncertainty can be decomposed into spatial uncertainty—resolved using SLAM—and uncertainty about the objects in the environment, formalized as the object scouting problem and modeled using the Locally Observable Markov Decision Process (LOMDP). We introduce a new planning framework specifically designed for object scouting with LOMDPs called the Scouting Partial-Order Planner (SPOP), which exploits the characteristics of partial order and regression planning to plan around knowledge gaps the robot may have about the existence, location, and state of relevant objects in its environment. Our results highlight the benefits of partial-order planning, demonstrating its suitability for object scouting due to its ability to identify absent but task-relevant objects, and show that it outperforms comparable planners in plan length, computation time, and execution time.

Extracting knowledge and reasoning from large language models (LLMs) offers a path to designing intelligent robots. Common approaches that leverage LLMs for planning are unable to recover when actions fail and resort to retrying failed actions without resolving the underlying cause. We propose a novel approach (CAPE) that generates corrective actions to resolve precondition errors during planning. CAPE improves the quality of generated plans through few-shot reasoning on action preconditions. Our approach enables embodied agents to execute more tasks than baseline methods while maintaining semantic correctness and minimizing re-prompting. In VirtualHome, CAPE improves a human-annotated plan correctness metric from 28.89% to 49.63% over SayCan, whilst achieving competitive executability. Our improvements transfer to a Boston Dynamics Spot robot initialized with a set of skills (specified in language) and associated preconditions, where CAPE improves correctness by 76.49% with higher executability compared to SayCan. Our approach enables embodied agents to follow natural language commands and robustly recover from failures.

Grounding spatiotemporal navigation commands to structured task specifications enables autonomous robots to understand a broad range of natural language and solve long-horizon tasks with safety guarantees. Prior works mostly focus on grounding spatial or temporally extended language for robots. We propose Lang2LTL-2, a modular system that leverages pretrained large language and vision-language models and multimodal semantic information to ground spatiotemporal navigation commands in novel city-scaled environments without retraining. Lang2LTL-2 achieves 93.53% language grounding accuracy on a dataset of 21,780 semantically diverse natural language commands in unseen environments. We run an ablation study to validate the need for different modalities. We also show that a physical robot equipped with the same system without modification can execute 50 semantically diverse natural language commands in both indoor and outdoor environments.

It is imperative that robots can understand natural language commands issued by humans. Such commands typically contain verbs that signify what action should be performed on a given object and that are applicable to many objects. We propose a method for generalizing manipulation skills to novel objects using verbs. Our method learns a probabilistic classifier that determines whether a given object trajectory can be described by a specific verb. We show that this classifier accurately generalizes to novel object categories with an average accuracy of 76.69% across 13 object categories and 14 verbs. We then perform policy search over the object kinematics to find an object trajectory that maximizes classifier prediction for a given verb. Our method allows a robot to generate a trajectory for a novel object based on a verb, which can then be used as input to a motion planner. We show that our model can generate trajectories that are usable for executing five verb commands applied to novel instances of two different object categories on a real robot.

In reality, there is still much to be done for robots to be able to perform manipulation actions with full autonomy. Complicated manipulation tasks, such as cooking, may still require a person to perform some actions that are very risky for a robot to perform. On the other hand, some other actions may be very risky for a human with physical disabilities to perform. Therefore, it is necessary to balance the workload of a robot and a human based on their limitations while minimizing the effort needed from a human in a collaborative robot (cobot) set-up. This paper proposes a new version of our functional object-oriented network (FOON) that integrates weights in its functional units to reflect a robot’s chance of successfully executing an action of that functional unit. The paper also presents a task planning algorithm for the weighted FOON to allocate manipulation action load to the robot and human to achieve optimal performance while minimizing human effort. Through a number of experiments, this paper shows several successful cases in which using the proposed weighted FOON and the task planning algorithm allow a robot and a human to successfully complete complicated tasks together with higher success rates than a robot doing them alone.

In this work, we propose a motion embedding strategy known as motion codes, which is a vectorized representation of motions based on a manipulation's salient mechanical attributes. These motion codes provide a robust motion representation, and they are obtained using a hierarchy of features called the motion taxonomy. We developed and trained a deep neural network model that combines visual and semantic features to identify the features found in our motion taxonomy to embed or annotate videos with motion codes. To demonstrate the potential of motion codes as features for machine learning tasks, we integrated the extracted features from the motion embedding model into the current state-of-the-art action recognition model. The obtained model achieved higher accuracy than the baseline model for the verb classification task on egocentric videos from the EPIC-KITCHENS dataset.

A motion taxonomy can encode manipulations as a binary-encoded representation, which we refer to as motion codes. These motion codes innately represent a manipulation action in an embedded space that describes the motion's mechanical features, including contact and trajectory type. The key advantage of using motion codes for embedding is that motions can be more appropriately defined with robotic-relevant features, and their distances can be more reasonably measured using these motion features. In this paper, we develop a deep learning pipeline to extract motion codes from demonstration videos in an unsupervised manner so that knowledge from these videos can be properly represented and used for robots. Our evaluations show that motion codes can be extracted from demonstrations of action in the EPIC-KITCHENS dataset.

This paper introduces a taxonomy of manipulations as seen especially in cooking for 1) grouping manipulations from the robotics point of view, 2) consolidating aliases and removing ambiguity for motion types, and 3) provide a path to transferring learned manipulations to new unlearned manipulations. Using instructional videos as a reference, we selected a list of common manipulation motions seen in cooking activities grouped into similar motions based on several trajectory and contact attributes. Manipulation codes are then developed based on the taxonomy attributes to represent the manipulation motions. The manipulation taxonomy is then used for comparing motion data in the Daily Interactive Manipulation (DIM) data set to reveal their motion similarities.

Video understanding is one of the most challenging topics in computer vision. In this paper, a four-stage video understanding pipeline is presented to simultaneously recognize all atomic actions and the single ongoing activity in a video. This pipeline uses objects and motions from the video and a graph-based knowledge representation network as prior reference. Two deep networks are trained to identify objects and motions in each video sequence associated with an action and low level image features are used to identify objects of interest in the video sequence. Confidence scores are assigned to objects of interest to represent their involvement in the action and to motion classes based on results from a deep neural network that classifies an ongoing action in video into motion classes. Confidence scores are computed for each candidate functional unit to associate them with an action using a knowledge representation network, object confidences, and motion confidences. Each action, therefore, is associated with a functional unit, and the sequence of actions is evaluated to identify the sole activity occurring in the video. The knowledge representation used in the pipeline is called the functional object-oriented network, which is a graph-based network useful for encoding knowledge about manipulation tasks. Experiments are performed on a dataset of cooking videos to test the proposed algorithm with action inference and activity classification. Experiments show that using a functional object-oriented network improves video understanding significantly.

We build upon the functional object-oriented network (FOON), a structured knowledge representation which is constructed from observations of human activities and manipulations. A FOON can be used for representing object-motion affordances. Knowledge retrieval through graph search allows us to obtain novel manipulation sequences using knowledge spanning across many video sources, hence the novelty in our approach. However, we are limited to the sources collected. To further improve the performance of knowledge retrieval as a follow up to our previous work, we discuss generalizing knowledge to be applied to objects which are similar to what we have in FOON without manually annotating new sources of knowledge. We discuss two means of generalization: 1) expanding our network through the use of object similarity to create new functional units from those we already have, and 2) compressing the functional units by object categories rather than specific objects. We discuss experiments which compare the performance of our knowledge retrieval algorithm with both expansion and compression by categories.

This paper presents a novel structured knowledge representation called the functional object-oriented network (FOON) to model the connectivity of the functional-related objects and their motions in manipulation tasks. The graphical model FOON is learned by observing object state change and human manipulations with the objects. Using a well-trained FOON, robots can decipher a task goal, seek the correct objects at the desired states on which to operate, and generate a sequence of proper manipulation motions. The paper describes FOON's structure and an approach to form a universal FOON with extracted knowledge from online instructional videos. A graph retrieval approach is presented to generate manipulation motion sequences from the FOON to achieve a desired goal, demonstrating the flexibility of FOON in creating a novel and adaptive means of solving a problem using knowledge gathered from multiple sources. The results are demonstrated in a simulated environment to illustrate the motion sequences generated from the FOON to carry out the desired tasks.