We focus on agile, continuous, and terrain-adaptive jumping of quadrupedal robots in discontinuous terrains such as stairs and stepping stones. Unlike single-step jumping, continuous jumping requires accurately executing highly dynamic motions over long horizons, which is challenging for existing approaches. To accomplish this task, we design a hierarchical learning and control framework, which consists of a learned heightmap predictor for robust terrain perception, a reinforcement-learning-based centroidal-level motion policy for versatile and terrain-adaptive planning, and a low-level model-based leg controller for accurate motion tracking. In addition, we minimize the sim-to-real gap by accurately modeling the hardware characteristics. Our framework enables a Unitree Go1 robot to perform agile and continuous jumps on human-sized stairs and sparse stepping stones, for the first time to the best of our knowledge. In particular, the robot can cross two stair steps in each jump and completes a 3.5m long, 2.8m high, 14-step staircase in 4.5 seconds. Moreover, the same policy outperforms baselines in various other parkour tasks, such as jumping over single horizontal or vertical discontinuities. Experiment videos can be found at https://yxyang.github.io/jumping_cod/.

Reliable estimation of terrain traversability is critical for the successful deployment of autonomous systems in wild, outdoor environments. Given the lack of large-scale annotated datasets for off-road navigation, strictly-supervised learning approaches remain limited in their generalization ability. To this end, we introduce a novel, image-based self-supervised learning method for traversability prediction, leveraging a state-of-the-art vision foundation model for improved out-of-distribution performance. Our method employs contrastive representation learning using both human driving data and instance-based segmentation masks during training. We show that this simple, yet effective, technique drastically outperforms recent methods in predicting traversability for both on- and off-trail driving scenarios. We compare our method with recent baselines on both a common benchmark as well as our own datasets, covering a diverse range of outdoor environments and varied terrain types. We also demonstrate the compatibility of resulting costmap predictions with a model-predictive controller. Finally, we evaluate our approach on zero- and few-shot tasks, demonstrating unprecedented performance for generalization to new environments. Videos and additional material can be found here: https://sites.google.com/view/visual-traversability-learning.

Improving long-term user satisfaction in user-centric networks while ensuring an acceptable delay in the presence of an eavesdropper is an important yet challenging task. In this paper, we investigate a long-term secure resource allocation for a user-centric network. Our objective is to maximize the user's long-term average satisfaction by controlling the network-layer data arrival rate and physical-layer power allocation at access points. This objective is mainly constrained by average delay requirements, average leakage rate, and data-queue stability conditions. These constraints are challenging to directly solve because of less-analytical expressions. To tackle this issue, we introduce two virtual queues as the debt of delay and leakage data. By using a Lyapunov optimization approach, we incorporate the stabilities of these virtual queues and the real data backlog queue into the objective function as a penalty. This allows us to transform the long-term optimization problem into a sequence of short-term subproblems that can be analytically solved at each slot. Simulation reveals that the proposed long-term secure resource allocation scheme outperforms the snapshot-based method in terms of the user's long-term satisfaction.

Efficient information gathering is a critical task for the Internet of underwater things (IoUT) that empowers the marine industry, but energy-constrained devices have become a bottleneck in the sustainability of IoUT. In this article, we focus on the unmanned aerial vehicle (UAV)-aided marine data collection in an air–ocean integrated network. We consider the mobility of buoys caused by waves during a collection period to achieve robust transmission, instead of assuming stationary buoys. Our objective is to reduce the energy consumption of UAV's flight and the communication energy of buoys and sensors via joint trajectory planning and resource allocation while ensuring transmission robustness. Due to the nonconvex objective function and coupled variables in constraints, we use an alternating optimization method to divide the original problem into two subproblems and solve them iteratively. The successive convex approximation technique is used to tackle the nonconvex constraints. Furthermore, the wave-induced mobility of buoys results in infinite number of constraints. To tackle this issue, we use an$\mathcal {S}$-procedure to transform these constraints into a series of deterministic linear matrix inequalities. Simulations reveal that the proposed scheme reduces energy consumption while achieving robust transmission. Specifically, the proposed scheme overcomes wave effects and achieves higher throughput than other nonrobust schemes. Meanwhile, our scheme is able to provide 27.3$\%$lower weighted energy consumption than the other robust transmission scheme.

We introduce LiDAR-UDA, a novel two-stage self-training-based Unsupervised Domain Adaptation (UDA) method for LiDAR segmentation. Existing self-training methods use a model trained on labeled source data to generate pseudo labels for target data and refine the predictions via fine-tuning the network on the pseudo labels. These methods suffer from domain shifts caused by different LiDAR sensor configurations in the source and target domains. We propose two techniques to reduce sensor discrepancy and improve pseudo label quality: 1) Li-DAR beam subsampling, which simulates different LiDAR scanning patterns by randomly dropping beams; 2) cross-frame ensembling, which exploits temporal consistency of consecutive frames to generate more reliable pseudo labels. Our method is simple, generalizable, and does not incur any extra inference cost. We evaluate our method on several public LiDAR datasets and show that it outperforms the state-of-the-art methods by more than 3.9% mIoU on average for all scenarios. Code will be available at https://github.com/JHLee0513/lidar_uda.

The current Internet of Underwater Things (IoUT) for marine observations and emergency responses suffers from two critical issues: 1) energy efficient and 2) timely data collection. Autonomous underwater vehicles (AUVs), serving as tools for collecting and forwarding distributed data, can deal with the unbalanced power consumption in a traditional multihop underwater communication network. However, the low speed of the AUV has not been able to guarantee the timeliness of delay-sensitive data. In this article, we introduce a hybrid data collection scheme (HDCS), taking both real-time data collection and energy efficiency (EE) issues into consideration. All sensor nodes (SNs) are first clustered based on their locations in the network. We develop an analytic expression to describe the attenuation of Value of Information (VoI), involving the relationship between the importance degree and timeliness; initial VoI could be measured by historical data. The emergency can be recognized by the presented criterion, and the transmission mode of cluster heads (CHs) in the same layer is judged by CHs themselves according to VoI. The selected CHs shall transmit the urgent data via multihop routing to avoid over attenuation of VoI. The normal data are collected by AUVs visiting all remaining CHs, and the shortest trajectory is achieved by addressing a variation of the classic traveling salesman problem (TSP). Our simulation experiments show that this mechanism can effectively increase long-term VoI while significantly improving EE.

The Internet of Underwater Things (IoUT) plays a key role in maritime monitoring systems, but energy-efficient data-uploading has been a challenging task owing to energy-constrained and expensive facilities, such as buoys and underwater sensors. In this article, we present an energy-efficient data collection scheme for unmanned aerial vehicle (UAV)-aided ocean monitoring networks (OMNs), where underwater acoustic and aerial radio frequency (RF) links are considered collaboratively. Our goal is to maximize energy efficiency (EE) of the entire OMN by jointly optimizing the transmit power of buoys and sensors, scheduling their transmissions, as well as designing the UAV's trajectory; the objective function is constrained by minimum throughput thresholds, power consumption budgets, and the UAV's kinematic conditions. Furthermore, we introduce a tradeoff between the energy consumption of buoys and sensors to bridge the gap between acoustic and RF links. The formulated problem is decomposed into three subproblems and they are solved alternatively. In each iteration, we leverage Dinkelbach's method and successive convex approximation (SCA) technique to tackle the fractional program (FP) and transform an original subproblem into a convex form, respectively. Extensive simulations confirm the convergence of our proposed scheme, reveal the influence of the tradeoff on EE, and show that our scheme outweighs other benchmarks in different scenarios.

Visual topological navigation has been revitalized recently thanks to the advancement of deep learning that substantially improves robot perception. However, the scalability and reliability issue remain challenging due to the complexity and ambiguity of real world images and mechanical constraints of real robots. We present an intuitive approach to show that by accurately measuring the capability of a local controller, large-scale visual topological navigation can be achieved while being scalable and robust. Our approach achieves state-of-the-art results in trajectory following and planning in large-scale environments. It also generalizes well to real robots and new environments without retraining or finetuning.

End-to-end learning for autonomous navigation has received substantial attention recently as a promising method for reducing modeling error. However, its data complexity, especially around generalization to unseen environments, is high. We introduce a novel image-based autonomous navigation technique that leverages in policy structure using the Riemannian Motion Policy (RMP) framework for deep learning of vehicular control. We design a deep neural network to predict control point RMPs of the vehicle from visual images, from which the optimal control commands can be computed analytically. We show that our network trained in the Gibson environment can be used for indoor obstacle avoidance and navigation on a real RC car, and our RMP representation generalizes better to unseen environments than predicting local geometry or predicting control commands directly.

Duty-cycling is generally adopted in existing sensor networks to reduce power consumption and these networks depend on neighbor discovery protocols to ensure that nodes wake up and discover each other. For different neighbor discovery protocols, the discovery latency is determined by two factors: the wake-sleep pattern and slot size. To the best of our knowledge, previous works on neighbor discovery have thus far been focused on improving the wake-sleep pattern. In this paper, we investigate the extent to which we can improve discovery latency by reducing the slot size. We found that by reducing the slot size, i.e., reducing the listening time in active slots, the collisions between beacons and synchronization between nodes become more severe, which can lead to discovery failures that are not predicted by existing theoretical models. We show that we can mitigate these effects by reducing the number of beacons and introducing randomization. We propose a new continuous-listening-based neighbor discovery algorithm called Spotlight. Our evaluations with a practical sensor testbed suggest that Spotlight can achieve a 50% reduction in discovery latency over existing state-of-the-art neighbor discovery protocols without increasing power consumption in existing sensor networks.

Recent advances in unmanned aerial vehicle (UAV) technologies have made it possible to deploy an aerial video surveillance system to provide an unprecedented aerial perspective for ground monitoring in real time. Multiple UAVs would be required to cover a large target area, and it is difficult for users to visualize the overall situation if they were to receive multiple disjoint video streams. To address this problem, we designed and implemented SkyStitch, a multiple-UAV video surveillance system that provides a single and panoramic video stream to its users by stitching together multiple aerial video streams. SkyStitch addresses two key design challenges: (i) the high computational cost of stitching and (ii) the difficulty of ensuring good stitching quality under dynamic conditions. To improve the speed and quality of video stitching, we incorporate several practical techniques like distributed feature extraction to reduce workload at the ground station, the use of hints from the flight controller to improve stitching efficiency and a Kalman filter-based state estimation model to mitigate jerkiness. Our results show that SkyStitch can achieve a stitching rate that is 4 times faster than existing state-of-the-art methods and also improve perceptual stitching quality. We also show that SkyStitch can be easily implemented using commercial off-the-shelf hardware.