We introduce PhotoBot, a framework for automated photo acquisition based on an interplay between high-level human language guidance and a robot photographer. We propose to communicate photography suggestions to the user via a reference picture that is retrieved from a curated gallery. We exploit a visual language model (VLM) and an object detector to characterize reference pictures via textual descriptions and use a large language model (LLM) to retrieve relevant reference pictures based on a user's language query through text-based reasoning. To correspond the reference picture and the observed scene, we exploit pre-trained features from a vision transformer capable of capturing semantic similarity across significantly varying images. Using these features, we compute pose adjustments for an RGB-D camera by solving a Perspective-n-Point (PnP) problem. We demonstrate our approach on a real-world manipulator equipped with a wrist camera. Our user studies show that photos taken by PhotoBot are often more aesthetically pleasing than those taken by users themselves, as measured by human feedback.

Abstract--Kinesthetic Teaching is a popular approach to collecting expert robotic demonstrations of contact-rich tasks for imitation learning (IL), but it typically only measures motion, ignoring the force placed on the environment by the robot. Furthermore, contact-rich tasks require accurate sensing of both reaching and touching, which can be difficult to provide with conventional sensing modalities. We address these challenges with a See-Through-your-Skin (STS) visuotactile sensor, using the sensor both (i) as a measurement tool to improve kinesthetic teaching, and (ii) as a policy input in contact-rich door manipulation tasks. An STS sensor can be switched between visual and tactile modes by leveraging a semi-transparent surface and controllable lighting, allowing for both pre-contact visual sensing and during-contact tactile sensing with a single sensor. First, we propose tactile force matching, a methodology that enables a robot to match forces read during kinesthetic teaching using tactile signals. Second, we develop a policy that controls STS mode switching, allowing a policy to learn the appropriate moment to switch an STS from its visual to its tactile mode. Finally, we study multiple observation configurations to compare and contrast the value of visual and tactile data from an STS with visual data Figure 1: Our STS sensor before and during contact with a cabinet knob from a wrist-mounted eye-in-hand camera. In visual mode, the camera sees through episodes from real-world manipulation experiments, we find that the gel and allows finding and reaching the knob, while tactile mode the inclusion of force matching raises average policy success rates provides contact-based feedback, via gel deformation and resultant by 62.5%, STS mode switching by 30.3%, and STS data as a dot displacement, upon initial contact and during opening. This dot policy input by 42.5%. Our results highlight the utility of seethrough displacement can also be used to measure a signal linearly related to tactile sensing for IL, both for data collection to allow force. Red circles highlight knob in sensor view.

We present Learning to Place by Picking (LPP), a method capable of autonomously collecting demonstrations for a family of placing tasks in which objects must be manipulated to specific locations. With LPP, we approach the learning of robotic object placement policies by reversing the grasping process and exploiting the inherent symmetry of the pick and place problems. Specifically, we obtain placing demonstrations from a set of grasp sequences of objects that are initially located at their target placement locations. Our system is capable of collecting hundreds of demonstrations without human intervention by using a combination of tactile sensing and compliant control for grasps. We train a policy directly from visual observations through behaviour cloning, using the autonomously-collected demonstrations. By doing so, the policy can generalize to object placement scenarios outside of the training environment without privileged information (e.g., placing a plate picked up from a table and not at the original placement location). We validate our approach on home robotic scenarios that include dishwasher loading and table setting. Our approach yields robotic placing policies that outperform policies trained with kinesthetic teaching, both in terms of performance and data efficiency, while requiring no human supervision.

Quickly and reliably finding accurate inverse kinematics (IK) solutions remains a challenging problem for robotic manipulation. Existing numerical solvers typically produce a single solution only and rely on local search techniques to minimize a highly nonconvex objective function. Recently, learning-based approaches that approximate the entire feasible set of solutions have shown promise as a means to generate multiple fast and accurate IK results in parallel. However, existing learning-based techniques have a significant drawback: each robot of interest requires a specialized model that must be trained from scratch. To address this shortcoming, we investigate a novel distance-geometric robot representation coupled with a graph structure that allows us to leverage the flexibility of graph neural networks (GNNs). We use this approach to train a generative graphical inverse kinematics solver (GGIK) that is able to produce a large number of diverse solutions in parallel while also generalizing well -- a single learned model can be used to produce IK solutions for a variety of different robots. The graphical formulation elegantly exposes the symmetry and Euclidean equivariance of the IK problem that stems from the spatial nature of robot manipulators. We exploit this symmetry by encoding it into the architecture of our learned model, yielding a flexible solver that is able to produce sets of IK solutions for multiple robots.

Quickly and reliably finding accurate inverse kinematics (IK) solutions remains a challenging problem for robotic manipulation. Existing numerical solvers are broadly applicable, but typically only produce a single solution and rely on local search techniques to minimize highly nonconvex objective functions. More recent learning-based approaches that approximate the entire feasible set of solutions have shown promise as a means to generate multiple fast and accurate IK results in parallel. However, existing learning-based techniques have a significant drawback: each robot of interest requires a specialized model that must be trained from scratch. To address this key shortcoming, we investigate a novel distance-geometric robot representation coupled with a graph structure that allows us to leverage the flexibility of graph neural networks (GNNs). We use this approach to train the first learned generative graphical inverse kinematics (GGIK) solver that is able to produce a large number of diverse solutions in parallel and to also generalize: a single learned model can be used to produce IK solutions for a variety of different robots. When compared to several other learned IK methods, GGIK provides more accurate solutions. GGIK is also able to generalize reasonably well to robot manipulators unseen during training. Finally, we show that GGIK can be used to complement local IK solvers by providing reliable initializations to seed the local optimization process.

Our work examines the way in which large language models can be used for robotic planning and sampling, specifically the context of automated photographic documentation. Specifically, we illustrate how to produce a photo-taking robot with an exceptional level of semantic awareness by leveraging recent advances in general purpose language (LM) and vision-language (VLM) models. Given a high-level description of an event we use an LM to generate a natural-language list of photo descriptions that one would expect a photographer to capture at the event. We then use a VLM to identify the best matches to these descriptions in the robot's video stream. The photo portfolios generated by our method are consistently rated as more appropriate to the event by human evaluators than those generated by existing methods.

Sequential modelling of high-dimensional data is an important problem that appears in many domains including model-based reinforcement learning and dynamics identification for control. Latent variable models applied to sequential data (i.e., latent dynamics models) have been shown to be a particularly effective probabilistic approach to solve this problem, especially when dealing with images. However, in many application areas (e.g., robotics), information from multiple sensing modalities is available -- existing latent dynamics methods have not yet been extended to effectively make use of such multimodal sequential data. Multimodal sensor streams can be correlated in a useful manner and often contain complementary information across modalities. In this work, we present a self-supervised generative modelling framework to jointly learn a probabilistic latent state representation of multimodal data and the respective dynamics. Using synthetic and real-world datasets from a multimodal robotic planar pushing task, we demonstrate that our approach leads to significant improvements in prediction and representation quality. Furthermore, we compare to the common learning baseline of concatenating each modality in the latent space and show that our principled probabilistic formulation performs better. Finally, despite being fully self-supervised, we demonstrate that our method is nearly as effective as an existing supervised approach that relies on ground truth labels.