Abstract
Robust planning in interactive scenarios requires predicting the uncertain future to make risk-aware decisions. Unfortunately, due to long-tail safety-critical events, the risk is often under-estimated by finite-sampling approximations of probabilistic motion forecasts. This can lead to overconfident and unsafe robot behavior, even with robust planners. Instead of assuming full prediction coverage that robust planners require, we propose to make prediction itself risk-aware. We introduce a new prediction objective to learn a risk-biased distribution over trajectories, so that risk evaluation simplifies to an expected cost estimation under this biased distribution. This reduces the sample complexity of the risk estimation during online planning, which is needed for safe real-time performance. Evaluation results in a didactic simulation environment and on a real-world dataset demonstrate the effectiveness of our approach.
Keyword: overconfidence
There is no result
Keyword: confidence
Online Update of Safety Assurances Using Confidence-Based Predictions
Abstract
Robots such as autonomous vehicles and assistive manipulators are increasingly operating in dynamic environments and close physical proximity to people. In such scenarios, the robot can leverage a human motion predictor to predict their future states and plan safe and efficient trajectories. However, no model is ever perfect -- when the observed human behavior deviates from the model predictions, the robot might plan unsafe maneuvers. Recent works have explored maintaining a confidence parameter in the human model to overcome this challenge, wherein the predicted human actions are tempered online based on the likelihood of the observed human action under the prediction model. This has opened up a new research challenge, i.e., \textit{how to compute the future human states online as the confidence parameter changes?} In this work, we propose a Hamilton-Jacobi (HJ) reachability-based approach to overcome this challenge. Treating the confidence parameter as a virtual state in the system, we compute a parameter-conditioned forward reachable tube (FRT) that provides the future human states as a function of the confidence parameter. Online, as the confidence parameter changes, we can simply query the corresponding FRT, and use it to update the robot plan. Computing parameter-conditioned FRT corresponds to an (offline) high-dimensional reachability problem, which we solve by leveraging recent advances in data-driven reachability analysis. Overall, our framework enables online maintenance and updates of safety assurances in human-robot interaction scenarios, even when the human prediction model is incorrect. We demonstrate our approach in several safety-critical autonomous driving scenarios, involving a state-of-the-art deep learning-based prediction model.
Data-Efficient Characterization of the Global Dynamics of Robot Controllers with Confidence Guarantees
Authors: Ewerton R. Vieira, Aravind Sivaramakrishnan, Yao Song, Edgar Granados, Marcio Gameiro, Konstantin Mischaikow, Ying Hung, Kostas E. Bekris
Abstract
This paper proposes an integration of surrogate modeling and topology to significantly reduce the amount of data required to describe the underlying global dynamics of robot controllers, including closed-box ones. A Gaussian Process (GP), trained with randomized short trajectories over the state-space, acts as a surrogate model for the underlying dynamical system. Then, a combinatorial representation is built and used to describe the dynamics in the form of a directed acyclic graph, known as {\it Morse graph}. The Morse graph is able to describe the system's attractors and their corresponding regions of attraction (\roa). Furthermore, a pointwise confidence level of the global dynamics estimation over the entire state space is provided. In contrast to alternatives, the framework does not require estimation of Lyapunov functions, alleviating the need for high prediction accuracy of the GP. The framework is suitable for data-driven controllers that do not expose an analytical model as long as Lipschitz-continuity is satisfied. The method is compared against established analytical and recent machine learning alternatives for estimating \roa s, outperforming them in data efficiency without sacrificing accuracy. Link to code: https://go.rutgers.edu/49hy35en
Safety-Aware Learning-Based Control of Systems with Uncertainty Dependent Constraints (extended version)
Abstract
The problem of safely learning and controlling a dynamical system - i.e., of stabilizing an originally (partially) unknown system while ensuring that it does not leave a prescribed 'safe set' - has recently received tremendous attention in the controls community. Further complexities arise, however, when the structure of the safe set itself depends on the unknown part of the system's dynamics. In particular, a popular approach based on control Lyapunov functions (CLF), control barrier functions (CBF) and Gaussian processes (to build confidence set around the unknown term), which has proved successful in the known-safe set setting, becomes inefficient as-is, due to the introduction of higher-order terms to be estimated and bounded with high probability using only system state measurements. In this paper, we build on the recent literature on GPs and reproducing kernels to perform this latter task, and show how to correspondingly modify the CLF-CBF-based approach to obtain safety guarantees. Namely, we derive exponential CLF and second relative order exponential CBF constraints whose satisfaction guarantees stability and forward in-variance of the partially unknown safe set with high probability. To overcome the intractability of verification of these conditions on the continuous domain, we apply discretization of the state space and use Lipschitz continuity properties of dynamics to derive equivalent CLF and CBF certificates in discrete state space. Finally, we present an algorithm for the control design aim using the derived certificates.
Keyword: scaling
Optimizing Data Collection for Machine Learning
Authors: Rafid Mahmood, James Lucas, Jose M. Alvarez, Sanja Fidler, Marc T. Law
Abstract
Modern deep learning systems require huge data sets to achieve impressive performance, but there is little guidance on how much or what kind of data to collect. Over-collecting data incurs unnecessary present costs, while under-collecting may incur future costs and delay workflows. We propose a new paradigm for modeling the data collection workflow as a formal optimal data collection problem that allows designers to specify performance targets, collection costs, a time horizon, and penalties for failing to meet the targets. Additionally, this formulation generalizes to tasks requiring multiple data sources, such as labeled and unlabeled data used in semi-supervised learning. To solve our problem, we develop Learn-Optimize-Collect (LOC), which minimizes expected future collection costs. Finally, we numerically compare our framework to the conventional baseline of estimating data requirements by extrapolating from neural scaling laws. We significantly reduce the risks of failing to meet desired performance targets on several classification, segmentation, and detection tasks, while maintaining low total collection costs.
How deep convolutional neural networks lose spatial information with training
Authors: Umberto M. Tomasini, Leonardo Petrini, Francesco Cagnetta, Matthieu Wyart
Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Abstract
A central question of machine learning is how deep nets manage to learn tasks in high dimensions. An appealing hypothesis is that they achieve this feat by building a representation of the data where information irrelevant to the task is lost. For image datasets, this view is supported by the observation that after (and not before) training, the neural representation becomes less and less sensitive to diffeomorphisms acting on images as the signal propagates through the net. This loss of sensitivity correlates with performance, and surprisingly correlates with a gain of sensitivity to white noise acquired during training. These facts are unexplained, and as we demonstrate still hold when white noise is added to the images of the training set. Here, we (i) show empirically for various architectures that stability to image diffeomorphisms is achieved by spatial pooling in the first half of the net, and by channel pooling in the second half, (ii) introduce a scale-detection task for a simple model of data where pooling is learned during training, which captures all empirical observations above and (iii) compute in this model how stability to diffeomorphisms and noise scale with depth. The scalings are found to depend on the presence of strides in the net architecture. We find that the increased sensitivity to noise is due to the perturbing noise piling up during pooling, after being rectified by ReLU units.
Monte Carlo Tree Search based Variable Selection for High Dimensional Bayesian Optimization
Authors: Lei Song, Ke Xue, Xiaobin Huang, Chao Qian
Abstract
Bayesian optimization (BO) is a class of popular methods for expensive black-box optimization, and has been widely applied to many scenarios. However, BO suffers from the curse of dimensionality, and scaling it to high-dimensional problems is still a challenge. In this paper, we propose a variable selection method MCTS-VS based on Monte Carlo tree search (MCTS), to iteratively select and optimize a subset of variables. That is, MCTS-VS constructs a low-dimensional subspace via MCTS and optimizes in the subspace with any BO algorithm. We give a theoretical analysis of the general variable selection method to reveal how it can work. Experiments on high-dimensional synthetic functions and real-world problems (i.e., NAS-bench problems and MuJoCo locomotion tasks) show that MCTS-VS equipped with a proper BO optimizer can achieve state-of-the-art performance.
Keyword: calibration
Force-Aware Interface via Electromyography for Natural VR/AR Interaction
Authors: Yunxiang Zhang, Benjamin Liang, Boyuan Chen, Paul Torrens, S. Farokh Atashzar, Dahua Lin, Qi Sun
Abstract
While tremendous advances in visual and auditory realism have been made for virtual and augmented reality (VR/AR), introducing a plausible sense of physicality into the virtual world remains challenging. Closing the gap between real-world physicality and immersive virtual experience requires a closed interaction loop: applying user-exerted physical forces to the virtual environment and generating haptic sensations back to the users. However, existing VR/AR solutions either completely ignore the force inputs from the users or rely on obtrusive sensing devices that compromise user experience. By identifying users' muscle activation patterns while engaging in VR/AR, we design a learning-based neural interface for natural and intuitive force inputs. Specifically, we show that lightweight electromyography sensors, resting non-invasively on users' forearm skin, inform and establish a robust understanding of their complex hand activities. Fuelled by a neural-network-based model, our interface can decode finger-wise forces in real-time with 3.3% mean error, and generalize to new users with little calibration. Through an interactive psychophysical study, we show that human perception of virtual objects' physical properties, such as stiffness, can be significantly enhanced by our interface. We further demonstrate that our interface enables ubiquitous control via finger tapping. Ultimately, we envision our findings to push forward research towards more realistic physicality in future VR/AR.
Open-source High-precision Autonomous Suturing Framework With Visual Guidance
Authors: Hongbin Lin, Bin Li, Yunhui Liu, Kwok Wai Samuel Au
Abstract
Autonomous surgery has attracted increasing attention for revolutionizing robotic patient care, yet remains a distant and challenging goal. In this paper, we propose an image-based framework for high-precision autonomous suturing operation. We first build an algebraic geometric algorithm to achieve accurate needle pose estimation, then design the corresponding keypoint-based calibration network for joint-offset compensation, and further plan and control suture trajectory. Our solution ranked first among all competitors in the AccelNet Surgical Robotics Challenge. The source code is opened here to accelerate future autonomous surgery research.
Non-learning Stereo-aided Depth Completion under Mis-projection via Selective Stereo Matching
Authors: Yasuhiro Yao, Ryoichi Ishikawa, Shingo Ando, Kana Kurata, Naoki Ito, Jun Shimamura, Takeshi Oishi
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
We propose a non-learning depth completion method for a sparse depth map captured using a light detection and ranging (LiDAR) sensor guided by a pair of stereo images. Generally, conventional stereo-aided depth completion methods have two limiations. (i) They assume the given sparse depth map is accurately aligned to the input image, whereas the alignment is difficult to achieve in practice. (ii) They have limited accuracy in the long range because the depth is estimated by pixel disparity. To solve the abovementioned limitations, we propose selective stereo matching (SSM) that searches the most appropriate depth value for each image pixel from its neighborly projected LiDAR points based on an energy minimization framework. This depth selection approach can handle any type of mis-projection. Moreover, SSM has an advantage in terms of long-range depth accuracy because it directly uses the LiDAR measurement rather than the depth acquired from the stereo. SSM is a discrete process; thus, we apply variational smoothing with binary anisotropic diffusion tensor (B-ADT) to generate a continuous depth map while preserving depth discontinuity across object boundaries. Experimentally, compared with the previous state-of-the-art stereo-aided depth completion, the proposed method reduced the mean absolute error (MAE) of the depth estimation to 0.65 times and demonstrated approximately twice more accurate estimation in the long range. Moreover, under various LiDAR-camera calibration errors, the proposed method reduced the depth estimation MAE to 0.34-0.93 times from previous depth completion methods.
Neural Implicit Surface Reconstruction from Noisy Camera Observations
Authors: Sarthak Gupta, Patrik Huber
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Representing 3D objects and scenes with neural radiance fields has become very popular over the last years. Recently, surface-based representations have been proposed, that allow to reconstruct 3D objects from simple photographs. However, most current techniques require an accurate camera calibration, i.e. camera parameters corresponding to each image, which is often a difficult task to do in real-life situations. To this end, we propose a method for learning 3D surfaces from noisy camera parameters. We show that we can learn camera parameters together with learning the surface representation, and demonstrate good quality 3D surface reconstruction even with noisy camera observations.
Keyword: out of distribution detection
There is no result
Keyword: out-of-distribution detection
There is no result
Keyword: expected calibration error
There is no result
Keyword: overconfident
RAP: Risk-Aware Prediction for Robust Planning
Keyword: overconfidence
There is no result
Keyword: confidence
Online Update of Safety Assurances Using Confidence-Based Predictions
Data-Efficient Characterization of the Global Dynamics of Robot Controllers with Confidence Guarantees
Safety-Aware Learning-Based Control of Systems with Uncertainty Dependent Constraints (extended version)
Keyword: scaling
Optimizing Data Collection for Machine Learning
How deep convolutional neural networks lose spatial information with training
Monte Carlo Tree Search based Variable Selection for High Dimensional Bayesian Optimization
Keyword: calibration
Force-Aware Interface via Electromyography for Natural VR/AR Interaction
Open-source High-precision Autonomous Suturing Framework With Visual Guidance
Non-learning Stereo-aided Depth Completion under Mis-projection via Selective Stereo Matching
Neural Implicit Surface Reconstruction from Noisy Camera Observations