Commercial plant phenotyping systems using fixed cameras cannot perceive many plant details due to leaf occlusion. In this paper, we present Botany-Bot, a system for building detailed “annotated digital twins” of living plants using two stereo cameras, a digital turntable inside a lightbox, an industrial robot arm, and 3D segmentated Gaussian Splat models. We also present robot algorithms for manipulating leaves to take high-resolution indexable images of occluded details such as stem buds and the underside/topside of leaves. Results from experiments suggest that Botany-Bot can segment leaves with 90.8% accuracy, detect leaves with 86.2% accuracy, lift/push leaves with 77.9% accuracy, and take detailed overside/underside images with 77.3% accuracy. Code, videos, and datasets are available at https://berkeleyautomation.github.io/Botany-Bot/.

A robotic behavior model that can reliably generate behaviors from natural language inputs in real time would substantially expedite the adoption of industrial robots due to enhanced system flexibility. To facilitate these efforts, we construct a framework in which learned behaviors, created by a natural language abstractor, are verifiable by construction. Leveraging recent advancements in motion primitives and probabilistic verification, we construct a natural-language behavior abstractor that generates behaviors by synthesizing a directed graph over the provided motion primitives. If these component motion primitives are constructed according to the criteria we specify, the resulting behaviors are probabilistically verifiable. We demonstrate this verifiable behavior generation capacity in both simulation on an exploration task and on hardware with a robot scooping granular media.

Polyculture farming has environmental advantages but requires substantially more labor than monoculture farming. We present novel hardware and algorithms for automated pruning and irrigation. Using an overhead camera to collect data from physical$1.5~m^{2}$garden testbeds, the autonomous system utilizes a learned Plant Phenotyping convolutional neural network and a Bounding Disk Tracking algorithm to evaluate the individual plant distribution and estimate the state of the garden each day. From this garden state, AlphaGardenSim selects plants to autonomously prune. A trained neural network detects and targets specific prune points on the plant. Two custom-designed pruning tools, compatible with a FarmBot commercial gantry system, are experimentally evaluated. Irrigation is automated using soil moisture sensors. We present results for four 60-day garden cycles. Results suggest the system can autonomously achieve 94% normalized plant diversity with pruning shears while maintaining an average canopy coverage of 84% by the end of the cycles. For code, videos, and datasets, see https://sites.google.com/berkeley.edu/pruningpolyculturej/home.Note to Practitioners—While polyculture farming is closer to how plants grow in nature, it is considered more labor intensive that monoculture farming. In this paper we present approaches and custom hardware for automation of pruning and irrigation. Physical experiments suggest that automation can yield both high coverage and diversity.

The AlphaGarden is an automated testbed for indoor polyculture farming which combines a first-order plant simulator, a gantry robot, a seed planting algorithm, plant phenotyping and tracking algorithms, irrigation sensors and algorithms, and custom pruning tools and algorithms. In this paper, we systematically compare the performance of the AlphaGarden to professional horticulturalists on the staff of the UC Berkeley Oxford Tract Greenhouse. The humans and the machine tend side-by-side polyculture gardens with the same seed arrangement. We compare performance in terms of canopy coverage, plant diversity, and water consumption. Results from two 60-day cycles suggest that the automated AlphaGarden performs comparably to professional horticulturalists in terms of coverage and diversity, and reduces water consumption by as much as 44%. Code, videos, and datasets are available at https//sites.google.com/berkeley.edulsystematiccomparison

Learning-based methods in robotics hold the promise of generalization, but what can be done if a learned policy does not generalize to a new situation? In principle, if an agent can at least evaluate its own success (i.e., with a reward classifier that generalizes well even when the policy does not), it could actively practice the task and finetune the policy in this situation. We study this problem in the setting of industrial insertion tasks, such as inserting connectors in sockets and setting screws. Existing algorithms rely on precise localization of the connector or socket and carefully managed physical setups, such as assembly lines, to succeed at the task. But in unstructured environments such as homes or even some industrial settings, robots cannot rely on precise localization and may be tasked with previously unseen connectors. Offline reinforcement learning on a variety of connector insertion tasks is a potential solution, but what if the robot is tasked with inserting previously unseen connector? In such a scenario, we will still need methods that can robustly solve such tasks with online practice. One of the main observations we make in this work is that, with a suitable representation learning and domain generalization approach, it can be significantly easier for the reward function to generalize to a new but structurally similar task (e.g., inserting a new type of connector) than for the policy. This means that a learned reward function can be used to facilitate the finetuning of the robot's policy in situations where the policy fails to generalize in zero shot, but the reward function generalizes successfully. We show that such an approach can be instantiated in the real world, pretrained on 50 different connectors, and successfully finetuned to new connectors via the learned reward function. Videos and visualizations can be viewed at sites.google.com/view/learningonthejob

We consider a decluttering problem where multiple rigid convex polygonal objects rest in randomly placed positions and orientations on a planar surface and must be efficiently transported to a packing box using both single and multi-object grasps. Prior work considered frictionless multi-object grasping. In this paper, we introduce friction to increase the number of potential grasps for a given group of objects, and thus increase picks per hour. We train a neural network using real examples to plan robust multi-object grasps. In physical experiments, we find a 13.7% increase in success rate, a 1.6x increase in picks per hour, and a 6.3x decrease in grasp planning time compared to prior work on multi-object grasping. Compared to single-object grasping, we find a 3.1x increase in picks per hour.

While deep learning has enabled significant progress in designing general purpose robot grasping systems, there remain objects which still pose challenges for these systems. Recent work on Exploratory Grasping has formalized the problem of systematically exploring grasps on these adversarial objects and explored a multi-armed bandit model for identifying high-quality grasps on each object stable pose. However, these systems are still limited to exploring a small number or grasps on each object. We present Learned Efficient Grasp Sets (LEGS), an algorithm that efficiently explores thousands of possible grasps by maintaining small active sets of promising grasps and determining when it can stop exploring the object with high confidence. Experiments suggest that LEGS can identify a high-quality grasp more efficiently than prior algorithms which do not use active sets. In simulation experiments, we measure the gap between the success probability of the best grasp identified by LEGS, baselines, and the most-robust grasp (verified ground truth). After 3000 exploration steps, LEGS outperforms baseline algorithms on 10/14 and 25/39 objects on the Dex-Net Adversarial and EGAD! datasets respectively. We then evaluate LEGS in physical experiments; trials on 3 challenging objects suggest that LEGS converges to high-performing grasps significantly faster than baselines. See https://sites.google.com/view/LEGS-exp-grasping for supplemental material and videos.

Connector insertion and many other tasks commonly found in modern manufacturing settings involve complex contact dynamics and friction. Since it is difficult to capture related physical effects with first-order modeling, traditional control methods often result in brittle and inaccurate controllers, which have to be manually tuned. Reinforcement learning (RL) methods have been demonstrated to be capable of learning controllers in such environments from autonomous interaction with the environment, but running RL algorithms in the real world poses sample efficiency and safety challenges. Moreover, in practical real-world settings, we cannot assume access to perfect state information or dense reward signals. In this paper, we consider a variety of difficult industrial insertion tasks with visual inputs and different natural reward specifications, namely sparse rewards and goal images. We show that methods that combine RL with prior information, such as classical controllers or demonstrations, can solve these tasks from a reasonable amount of real-world interaction.

Robotic insertion tasks are characterized by contact and friction mechanics, making them challenging for conventional feedback control methods due to unmodeled physical effects. Reinforcement learning (RL) is a promising approach for learning control policies in such settings. However, RL can be unsafe during exploration and might require a large amount of real-world training data, which is expensive to collect. In this paper, we study how to use meta-reinforcement learning to solve the bulk of the problem in simulation by solving a family of simulated industrial insertion tasks and then adapt policies quickly in the real world. We demonstrate our approach by training an agent to successfully perform challenging real-world insertion tasks using less than 20 trials of real-world experience.

Navigation and control are a largely unsolved problems for micro autonomous underwater vehicles (μAUVs). The main challenges are due to the lack of accurate underwater localization systems, which fit on-board of μAUVs. In this work, we present an integrated navigation and control architecture consisting of a low-cost embedded localization module and an underwater way-point tracking controller, which fulfills the requirements of μAUVs. The performance of the navigation and control system is benchmarked in two different experimental scenarios.

Conventional feedback control methods can solve various types of robot control problems very efficiently by capturing the structure with explicit models, such as rigid body equations of motion. However, many control problems in modern manufacturing deal with contacts and friction, which are difficult to capture with first-order physical modeling. Hence, applying control design methodologies to these kinds of problems often results in brittle and inaccurate controllers, which have to be manually tuned for deployment. Reinforcement learning (RL) methods have been demonstrated to be capable of learning continuous robot controllers from interactions with the environment, even for problems that include friction and contacts. In this paper, we study how we can solve difficult control problems in the real world by decomposing them into a part that is solved efficiently by conventional feedback control methods, and the residual which is solved with RL. The final control policy is a superposition of both control signals. We demonstrate our approach by training an agent to successfully perform a real-world block assembly task involving contacts and unstable objects.

Precise robotic manipulation skills are desirable in many industrial settings, reinforcement learning (RL) methods hold the promise of acquiring these skills autonomously. In this paper, we explicitly consider incorporating operational space force/torque information into reinforcement learning; this is motivated by humans heuristically mapping perceived forces to control actions, which results in completing high-precision tasks in a fairly easy manner. Our approach combines RL with force/torque information by incorporating a proper operational space force controller; where we also exploit different ablations on processing this information. Moreover, we propose a neural network architecture that generalizes to reasonable variations of the environment. We evaluate our method on the open-source Siemens Robot Learning Challenge, which requires precise and delicate force-controlled behavior to assemble a tight-fit gear wheel set.

Accurate state estimation is a fundamental component of robotic control. In robotic manipulation tasks, as is our focus in this work, state estimation is essential for identifying the positions of objects in the scene, forming the basis of the manipulation plan. However, pose estimation typically requires expensive 3D cameras or additional instrumentation such as fiducial markers to perform accurately. Recently, Tobin et al. introduced an approach to pose estimation based on domain randomization, where a neural network is trained to predict pose directly from a 2D image of the scene. The network is trained on computer generated images with a high variation in textures and lighting, thereby generalizing to real world images. In this work, we investigate how to improve the accuracy of domain randomization based pose estimation. Our main idea is that active perception - moving the robot to get a better estimate of pose- can be trained in simulation and transferred to real using domain randomization. In our approach, the robot trains in a domain-randomized simulation how to estimate pose from a sequence of images. We show that our approach can significantly improve the accuracy of standard pose estimation in several scenarios: when the robot holding an object moves, when reference objects are moved in the scene, or when the camera is moved around the object.

Reinforcement learning (RL) allows robots to solve control tasks through interaction with their environment. In this paper we study a model-based value-function RL approach, which is suitable for computationally limited robots and light embedded systems. We develop a diving agent, which uses the RL algorithm for underwater depth stabilization. Simulations and experiments with the micro diving agent demonstrate its ability to learn the depth stabilization task.

We present the new HippoCampus micro underwater vehicle, first introduced in [1]. It is designed for monitoring confined fluid volumes. These tightly constrained settings demand agile vehicle dynamics. Moreover, we adapt a robust attitude control scheme for aerial drones to the underwater domain. We demonstrate the performance of the controller with a challenging maneuver. A submerged Furuta pendulum is stabilized by HippoCampus after a swing-up. The experimental results reveal the robustness of the control method, as the system quickly recovers from strong physical disturbances, which are applied to the system.

Reinforcement learning for assembly tasks can yield powerful robot control algorithms for applications that are challenging or even impossible for “conventional” feedback control methods. Insertion of a rigid peg into a deformable hole of smaller diameter is such a task. In this contribution we solve this task with Deep Reinforcement Learning. Force-torque measurements from a robot arm wrist sensor are thereby incorporated two-fold; they are integrated into the policy learning process and they are exploited in an admittance controller that is coupled to the neural network. This enables robot learning of contact-rich assembly tasks without explicit joint torque control or passive mechanical compliance. We demonstrate our approach in experiments with an industrial robot.

We present an approach for estimating the absolute poses of a swarm Micro Autonomous Underwater Vehicles (μAUVs) by decomposing the problem into few absolute position estimations and many relative pose estimations. As power constraints are critical to small mobile robots, we develop an extension of active marker pose estimation using color information to solve the marker correspondence problem, and show that this approach is more energy efficient than reflective estimation approaches. We show the feasibility of this approach by localizing a robot navigating in an underwater test tank environment. Detailed analysis is presented characterizing the noise and error properties when estimating robot poses from fixed on-board markers. Moreover, we provide comparisons in power and computational cost for other popular methods of underwater localization.

A hyperbolic acoustic system for underwater robot self-localization is presented. Anchored transducers send acoustic signals which are observed by a receiver. The system is passive with one-way signal transmission. Time differences of arrival (TDOAs) between the emitted signals are estimated by the receiver via cross-correlation. These TDOAs are fed to an Extended Kalman Filter to estimate the global position of the receiver. We describe the complete signal processing chain as well as challenges in hardware and software design. Experimental results in air and water show the feasibility of the system. This paper demonstrates that acoustic one-way localization is possible with off-the-shelf hardware in experimental test tanks.

The HippoCampus platform is a low-cost micro autonomous underwater vehicle for swarm robotics research. This paper presents the hardware and software design, the communication link, instrumentation, and control system. The quadrotor design enables the vehicle to perform agile maneuvers in a very confined test tank. Vehicle navigation is based on the on-board sensor suite. Autonomous path following is presented and experimental results are evaluated.

The development of a semi-automated haptic device for installing cables is a promising application for ship building industry, as well as in other industries which require power cable installations. The most critical tasks of this device involve quickly and reliably grasping and releasing cables. In this study, we design a device with an end-effector that uses a self-locking mechanism based on rope elements. Here, the pulling force is directly transferred to the cable using a frictional connection. The modules of this device are designed for one-person-operation, and optimize for the harsh environment of shipyards. A set of equations of motions were derived from a mathematical model of the connection between rope elements of the end effector and the cable, which show that the presented design provides the necessary no-slip condition between the end effector and the cable. A printed circuit board with an 8-bit microcontroller controls the induction motor which gives 0.6 horsepower, in combination with an AC motor driver. Thereby sensor signals are processed to ensure the desired functions by the operator's commands. This device has been shown to perform with enhanced efficiency in cable-installing tasks throughout the field test in a shipyard.