IROS 2022 Papers

In recent years, with the combination of tissue engineering and additive manufacturing technologies, the possibility of fabricating scaffolds with porosity and complex structure has been improved. Since the properties of most biomaterial inks are influenced by temperature and thereby affect the quality of the scaffolds, a controlled printing environment is very important. This study focuses on temperature monitoring from the nozzle to the working platform. A compact heating jacket is developed to heat the needle and sense its temperature inside the nozzle. It makes it very different from common cartridge heating mechanisms. Moreover, a semi-closed printing environment composed of an air curtain and temperature circulation device is developed to create a stable cooling environment. It improves the uniformity of the work platform and increases by 50% the cooling time efficiency. To demonstrate the robustness for a wide range of temperatures, this study presents two experiments of printing two biomaterial inks at body and low temperatures, respectively.

Autonomous robots are prone to fail in real world environments, where unknown factors can cause their world model to be inaccurate or incomplete. This causes robots to become stuck or repeatedly perform unsuccessful actions believing it is the best option. Static switching frameworks such as sequential composition guarantee stability of the overall system if its constituents are also stable. However, due to unknown factors, robot actions may fail when the robot senses the environment inaccurately. We propose Adaptive Sequential Composition, a novel framework that dynamically selects robot behaviours based on their utility to achieving success. We show the usefulness of the framework in a simulated second order system task as well as its application in navigating a robot through narrow gaps, where the clearance with the gap is smaller than position accuracy. Simulated results show adaptive sequential composition outperforms sequential composition by up to 30 % when presented with unknown factors leading to behaviour failure. For navigating through a narrow gap, adaptive sequential composition improved success by 65 %.

In this paper, we present an online adaptive planning strategy for a team of robots with heterogeneous sensors to sample from a latent spatial field using a learned model for decision making. Current robotic sampling methods seek to gather information about an observable spatial field. However, many applications, such as environmental monitoring and precision agriculture, involve phenomena that are not directly observable or are costly to measure, called latent phenomena. In our approach, we seek to reason about the latent phenomenon in real-time by effectively sampling the observable spatial fields using a team of robots with heterogeneous sensors, where each robot has a distinct sensor to measure a different observable field. The information gain is estimated using a learned model that maps from the observable spatial fields to the latent phenomenon. This model captures aleatoric uncertainty in the relationship to allow for information theoretic measures. Additionally, we explicitly consider the correlations among the observable spatial fields, capturing the relationship between sensor types whose observations are not independent. We show it is possible to learn these correlations, and investigate the impact of the learned correlation models on the performance of our sampling approach. Through our qualitative and quantitative results, we illustrate that empirically learned correlations improve the overall sampling efficiency of the team. We simulate our approach using a data set of sensor measurements collected on Lac Hertel, in Quebec, which we make publicly available.

Autonomously selecting the right sequence of locations to sample is critical during exploration missions in unknown environments, with constraints on the number of samples that can be collected, and a possibility of system failure. A key idea for decision-making in unknown environments is to exploit side information available to the agent, combined with the information gained from samples collected so far, to estimate the sampling values. In this paper, we pose the problem of sampling site selection as a problem of finding the optimal policy in a Markov decision process modeling the unknown sampling values and the outcomes associated with sampling attempts at different locations. Our solution exploits the fact that the partially unknown rewards of this Markov decision process are correlated to each other to devise a strategy that attempts to maximize the total sample value while also ensuring that the agent achieves its minimum mission requirement. We validate the utility of the proposed approach by evaluating the method against a baseline strategy that pursues collecting the samples that are estimated to be of the highest value. Our evaluations use a simulated sampling problem on Martian terrain and using OceanWATERS, a high-fidelity simulator of a future Europa lander mission.

In this paper, we present an approach that enables long-term monitoring of biological activity on coral reefs by extending mission time and adaptively focusing sensing resources on high-value periods. Coral reefs are one of the most biodiverse ecosystems on the planet; yet they are also among the most imperiled: facing bleaching, ecological community collapses due to global climate change, and degradation from human activities. Our proposed method improves the ability of scientists to monitor biological activity and abundance using passive acoustic sensors. We accomplish this by extracting periodicities from the observed abundance, and using them to predict future abundance. This predictive model is then used with a Monte Carlo Tree Search planning algorithm to schedule sampling at periods of high biological activity, and power down the sensor during periods of low activity. In simulated experiments using long-term acoustic datasets collected in the US Virgin Islands, our adaptive Online Sensor Scheduling algorithm is able to double the lifetime of a sensor while simultaneously increasing the average observed acoustic activity by 21%.

This paper proposes a novel extended Kalman filter (EKF) along with its adaptive variant for effective magnetic, angular rate and gravity (MARG) sensor-only pose estimation of mobile robots operated longer periods in reference-denied environments. First, a gradient-descent orientation-based EKF framework is derived, which formulates the MARG-based pose propagation with both bandpass-filtered and bias compensated external acceleration signals. The proposed approach uses two correction signals beside the orientation update, namely, virtual observations and sparse reference signals are incorporated in the state correction. Next, the instantaneous dynamics is characterized by accelerometer/gyroscope signals-based measures and an adaptive strategy is derived for real-time tuning of EKF parameters. The algorithm is fine tuned in an optimization framework on an appropriate database. This database of ground truth and raw MARG measurements contains 16 robot motion scenarios, where both slow motions and agile maneuvers are performed on different terrains. The conducted analysis highlights that the proposed algorithms outperform the standard approaches, moreover, the adaptive strategy further improves the performance by 13%. The comprehensive performance evaluation demonstrates the efficacy of the new algorithms, thereby these robust approaches are proposed in environments characterized by sparse reference measurements.

The major challenges of collision avoidance for robot navigation in crowded scenes lie in accurate environment modeling, fast perceptions, and trustworthy motion planning policies. This paper presents a novel adaptive environment model based collision avoidance reinforcement learning (i.e., AEMCARL) framework for an unmanned robot to achieve collision-free motions in challenging navigation scenarios. The novelty of this work is threefold: (1) developing a hierarchical network of gated-recurrent-unit (GRU) for environment modeling; (2) developing an adaptive perception mechanism with an attention module; (3) developing an adaptive reward function for the reinforcement learning (RL) framework to jointly train the environment model, perception function and motion planning policy. The proposed method is tested with the Gym-Gazebo simulator and a group of robots (Husky and Turtlebot) under various crowded scenes. Both simulation and experimental results have demonstrated the superior performance of the proposed method over baseline methods.

We present a method for solving the coverage problem with the objective of autonomously exploring an unknown environment under mission time constraints. Here, the robot is tasked with planning a path over a horizon such that the accumulated area swept out by its sensor footprint is maximized. Because this problem exhibits a diminishing returns property known as submodularity, we choose to formulate it as a tree-based sequential decision making process. This formulation allows us to evaluate the effects of the robot's actions on future world coverage states, while simultaneously accounting for traversability risk and the dynamic constraints of the robot. To quickly find near-optimal solutions, we propose an effective approximation to the coverage sensor model which adapts to the local environment. Our method was extensively tested across various complex environments and served as the local exploration algorithm for a competing entry in the DARPA Subterranean Challenge.

Recent advances in legged locomotion have en-abled quadrupeds to walk on challenging terrains. However, bipedal robots are inherently more unstable and hence it's harder to design walking controllers for them. In this work, we leverage recent advances in rapid adaptation for locomotion control, and extend them to work on bipedal robots. Similar to existing works, we start with a base policy which produces actions while taking as input an estimated extrinsics vector from an adaptation module. This extrinsics vector contains information about the environment and enables the walking controller to rapidly adapt online. However, the extrinsics estimator could be imperfect, which might lead to poor performance of the base policy which expects a perfect estimator. In this paper, we propose A-RMA (Adapting RMA), which additionally adapts the base policy for the imperfect extrinsics estimator by finetuning it using model-free RL. We demonstrate that A-RMA outperforms a number of RL-based baseline controllers and model-based controllers in simulation, and show zero-shot deployment of a single A-RMA policy to enable a bipedal robot, Cassie, to walk in a variety of different scenarios in the real world beyond what it has seen during training. Videos and results at https: //ashish-kmr.github.io/a-rma/

Tactile signals provide rich information about objects via touch and are essential for a robot to perform dex-terous manipulation. Exploring actively via tactile perception collects important information about the workspace. However, designing an effective tactile exploration policy is challenging in unstructured environments. Typically, the geometric information is incomplete, and need to be completed by actively and repeatedly interacting with the environment. In this paper, we address the tactile exploration problem by proposing a shape-information-dependent exploration strategy, which consists of two components: (1) a Shape-Belief Encoder that encodes the explored area by learning effective 3-D reconstruction and predicts the complete object shape; (2) a shape-dependent exploration policy which incorporates the encoding in (1) to plan an exploration trajectory. The policy actively acquires new information about object surface by executing exploration actions. The Shape-Belief Encoder leverages the newly collected contact points to update the surface model and guides future exploration. We validate the proposed algorithm on simulated and real robots.

This paper considers active SLAM problem for 3D deformable environments where the trajectory of the robot is planned to optimize the SLAM results. A planning strategy combining an efficient global planner with an accurate local planner is proposed to solve the problem. Simulation results under different scenarios have shown that the proposed active SLAM algorithm provides a good balance between accuracy and efficiency as compared to the local planner and the global planner. The MATLAB code of this first active SLAM algorithm for 3D deformable environments is made publicly available4.

The problem of active mapping aims to plan an informative sequence of sensing views given a limited budget such as distance traveled. This paper considers active occupancy grid mapping using a range sensor, such as LiDAR or depth camera. State-of-the-art methods optimize information-theoretic measures relating the occupancy grid probabilities with the range sensor measurements. The non-smooth nature of ray-tracing within a grid representation makes the objective function non-differentiable, forcing existing methods to search over a discrete space of candidate trajectories. This work proposes a differentiable approximation of the Shannon mutual information between a grid map and ray-based observations that enables gradient ascent optimization in the continuous space of SE(3) sensor poses. Our gradient-based formulation leads to more informative sensing trajectories, while avoiding occlusions and collisions. The proposed method is demonstrated in simulated and real-world experiments in 2-D and 3-D environments. Materials supplementing this paper are available at: https://arashasgharivaskasi-bc.github.io/grad_active_mapping/

Robotic manipulation stands as a largely unsolved problem despite significant advances in robotics and machine learning in recent years. One of the key challenges in manipulation is the exploration of the dynamics of the environment when there is continuous contact between the objects being manipulated. This paper proposes a model-based active exploration approach that enables efficient learning in sparse-reward robotic manipulation tasks. The proposed method estimates an information gain objective using an ensemble of probabilistic models and deploys model predictive control (MPC) to plan actions online that maximize the expected reward while also performing directed exploration. We evaluate our proposed algorithm in simulation and on a real robot, trained from scratch with our method, on a challenging ball pushing task on tilted tables, where the target ball position is not known to the agent a-priori. Our real-world robot experiment serves as a fundamental application of active exploration in model-based reinforcement learning of complex robotic manipulation tasks. Project page https://sites.google.com/view/aerm.

Having accurate localization capabilities is one of the fundamental requirements of autonomous robots. For underwater vehicles, the choices for effective localization are limited due to limitations of GPS use in water and poor environ-mental visibility that makes camera-based methods ineffective. Popular inertial navigation methods for underwater localization using Doppler-velocity log sensors, sonar, high-end inertial navigation systems, or acoustic positioning systems require bulky expensive hardware which are incompatible with low-cost, bio-inspired underwater robots. In this paper, we introduce an approach for underwater robot localization inspired by GPS methods known as acoustic pseudoranging. Our method allows us to potentially localize multiple bio-inspired robots equipped with commonly available micro electro-mechanical systems microphones. This is achieved through estimating the time difference of arrival of acoustic signals sent simultaneously through four speakers with a known constellation geometry. We also leverage the same acoustic framework to perform one-way communication with the robot to execute some primitive motions. To our knowledge, this is the first application of the approach for the on-board localization of small bio-inspired robots in water. Hardware schematics and the accompanying code are released to aid further development in the field33https://github.com/rpl-cmu/underwater-acoustic-pseudoranging.

Programming paths for robotic welding conventionally requires precise positioning of workpieces, detailed 3D models and/or tedious teach pendant programming. A new method is introduced in this paper that enables an operator to teach the weld path to the robot through a haptic-visual interface. The operator teaches the path by guiding the tool tip to contact on the workpiece surface with force feedback through the haptic device, and drawing exploratory paths that intersect the edge to be welded as well as adjoining surfaces. Tool-tip positions in contact with the workpiece are recorded. A RANSAC-type algorithm is used to automatically estimate a piecewise parametric curve along the edge as well as geometric parameters of the adjoining surfaces. The required tool trajectory for the robot to weld along the workpiece edge is automatically generated. Experiments performed in simulation and on a physical KUKA IIWA7 robot demonstrate that the developed method can successfully detect workpiece edges within a maximum deviation of 1mm. Furthermore, the method is intuitive, and requires no knowledge of robot programming for an operator to program multi-segment weld paths quickly.

Fixed-wing drones must navigate to the desired location accurately for maneuvers such as picking up objects and perching. However, current GNSS receivers limit their navigation accuracy to several meters in outdoor environments, making such maneuvers impossible. RTK GNSS can improve flight accuracy, but it requires ground stations at the target location and additional communication modules on the drone. Here, we describe a fixed-wing platform with onboard computation that uses positional information from a GNSS receiver and vision from an onboard camera. The drone relies on a GNSS signal for flying towards a point of interest and switches to vision-based information to accurately reach the target. We conducted outdoor experiments to compare the flight accuracy of three navigation methods: GNSS, RTK GNSS, and the proposed GNSS-vision method. We also systematically assessed the robustness of vision-based control to compensate for GNSS errors and quantify the accuracy of the proposed method. Our results show that the accuracy of the proposed GNSS-vision system is on par with RTK GNSS. GNSS-vision reduces the average error of GNSS by over an order of magnitude, from 3.033 m to 0.283 m, and reduces the variance across repeated flights from 2.095 m to 0.309 m. We open-source the software-hardware architecture used in this paper to enable the research community to build on these results and expand the capabilities of fixed-wing drones.

Robotic comanipulation provides a cost-effective solution to telesurgery when remote operation is not strictly necessary. Within the field of laparoscopic surgery, the comanip-ulation scenario is only recently being exploited commercially in the form of lightweight backdrivable systems. A passive wrist backdrivable robot does not require preoperative alignment with the incision that acts as a fulcrum around which the laparoscopic instrument pivots. Moreover, backdrivable systems can be comanipulated by the user without the need for expensive force sensors. Unfortunately, most backdrivable systems only provide limited accuracy when measuring the end effector pose from their joint encoders. Accurate knowledge of the end effector pose is required to estimate the the instrument tip and fulcrum position. This work presents a robust method to improve localisation of the pose of the end effector of a backdrivable robot. The method fuses optical tracking with robot proprioception by means of an unscented Kalman filter and is robust against intermittent occlusions of the line of sight. The algorithm is experimentally validated by analyzing its initialization behavior and accuracy when estimating the instrument tip and fulcrum position. An accuracy of 1.58\pm 0.1571.58\pm 0.157 mm and 0.699\pm 0.3890.699\pm 0.389 mm is achieved when estimating the instrument tip and fulcrum position respectively, which makes the algorithm suitable for advanced guidance schemes in comanipulation robotic surgery.

We present a new solution to the fine-grained retrieval of clean CAD models from a large-scale database in order to recover detailed object shape geometries for RGBD scans. Unlike previous work simply indexing into a moderately small database using an object shape descriptor and accepting the top retrieval result, we argue that in the case of a large-scale database a more accurate model may be found within a neighborhood of the descriptor. More importantly, we propose that the distinctiveness deficiency of shape descriptors at the instance level can be compensated by a geometry-based re-ranking of its neighborhood. Our approach first leverages the discriminative power of learned representations to distinguish between different categories of models and then uses a novel robust point set distance metric to re-rank the CAD neighbor-hood, enabling fine-grained retrieval in a large shape database. Evaluation on a real-world dataset shows that our geometry-based re-ranking is a conceptually simple but highly effective method that can lead to a significant improvement in retrieval accuracy compared to the state-of-the-art.

This paper addresses the problem of learning control policies for mobile robots, modeled as unknown Markov Decision Processes (MDPs), that are tasked with temporal logic missions, such as sequencing, coverage, or surveillance. The MDP captures uncertainty in the workspace structure and the outcomes of control decisions. The control objective is to synthesize a control policy that maximizes the probability of accomplishing a high-level task, specified as a Linear Temporal Logic (LTL) formula. To address this problem, we propose a novel accelerated model-based reinforcement learning (RL) algorithm for LTL control objectives that is capable of learning control policies significantly faster than related approaches. Its sample-efficiency relies on biasing exploration towards directions that may contribute to task satisfaction. This is accomplished by leveraging an automaton representation of the LTL task as well as a continuously learned MDP model. Finally, we provide comparative experiments that demonstrate the sample efficiency of the proposed method against recent RL methods for LTL objectives.