Underwater 3D scene reconstruction is crucial for underwater robotic perception and navigation. However, the task is significantly challenged by the complex interplay between light propagation, water medium, and object surfaces, with existing methods unable to model their interactions accurately. Additionally, expensive training and rendering costs limit their practical application in underwater robotic systems. Therefore, we propose T ensorized Underwater Gaussian Splatting (TUGS), which can effectively solve the modeling challenges of the complex interactions between object geometries and water media while achieving significant parameter reduction. TUGS employs lightweight tensorized higher-order Gaussians with a physics-based underwater Adaptive Medium Estimation (AME) module, enabling accurate simulation of both light attenuation and backscatter effects in underwater environments. Compared to other NeRF-based and GS-based methods designed for underwater, TUGS is able to render high-quality underwater images with faster rendering speeds and less memory usage. Extensive experiments on real-world underwater datasets have demonstrated that TUGS can efficiently achieve superior reconstruction quality using a limited number of parameters, making it particularly suitable for memory-constrained underwater UA V applications.

Seeing-eye robots are very useful tools for guiding visually impaired people, potentially producing a huge societal impact given the low availability and high cost of real guide dogs. Although a few seeing-eye robot systems have already been demonstrated, none considered external tugs from humans, which frequently occur in a real guide dog setting. In this paper, we simultaneously train a locomotion controller that is robust to external tugging forces via Reinforcement Learning (RL), and an external force estimator via supervised learning. The controller ensures stable walking, and the force estimator enables the robot to respond to the external forces from the human. These forces are used to guide the robot to the global goal, which is unknown to the robot, while the robot guides the human around nearby obstacles via a local planner. Experimental results in simulation and on hardware show that our controller is robust to external forces, and our seeing-eye system can accurately detect force direction. We demonstrate our full seeing-eye robot system on a real quadruped robot with a blindfolded human. The video can be seen at our project page: https://bu-air-lab.github.io/guide_dog/

We investigate a deep reinforcement learning (RL) architecture that supports explaining why a learned agent prefers one action over another. The key idea is to learn action-values that are directly represented via human-understandable properties of expected futures. This is realized via the embedded self-prediction (ESP)model, which learns said properties in terms of human provided features. Action preferences can then be explained by contrasting the future properties predicted for each action. To address cases where there are a large number of features, we develop a novel method for computing minimal sufficient explanations from anESP. Our case studies in three domains, including a complex strategy game, show that ESP models can be effectively learned and support insightful explanations.

A multitude of factors can contribute to a flight being delayed, but Air France, who partnered with a handful of other companies, is testing the world's first self-driving luggage tug in hopes of streamlining airport operations and improving the speed of getting luggage to and from an aircraft. The vehicle, known as the AT135 baggage tractor, began official testing at France's Toulouse-Blagnac airport last month on November 15. To the untrained eye it looks like the myriad of vehicles you already see scurrying around the airport tarmac while waiting for a flight, including a cab with a seat, steering wheel, and all the controls needed for a human driver. But look closer and you'll be able to spot some of the telltale hardware upgrades of an autonomous vehicle, including laser scanning LIDAR sensors on the roof and bumper that complement less visible sensors like GPS and front and rear cameras providing a 360-degree view around the tug. Climb inside the tug and you'll also find a big switch allowing it to be switched between manual and autonomous modes, as well as an oversized touchscreen showing a map of the airport and all the gates the vehicle is designed to service.

The pig looks like any other pig, only it's been wearing a backpack for a week--in the name of science. Just behind its head sits a control box, with a battery and processor, from which runs a cable that enters through the pig's flank. Once inside, the cable attaches to a very special robot clamped onto the pig's esophagus, the pathway to the stomach. Little by little, the robot lengthens, in turn lengthening the tube. The robot attached to a segment of esophagus.

Robots seem so far away. We're so many years from Jetsons-esque machines that live among us and wash our dishes and fold our clothes. But the reality is the robots have arrived--you're just not noticing them. Take a robot called Tug, for instance. No, Tug can't talk philosophy with you, and Tug can't do your laundry.

We introduce an application of self-driving vehicle technology to the problem of towing aircraft at busy airports from gate to runway and runway to gate. Autonomous towing can be supervised by human ramp- or ATC controllers, pilots, or ground crew. The controllers provide route information to the tugs, assisted by an automated route planning system. The planning system and tower and ground controllers work in conjunction with the tugs to make tactical decisions during operations to ensure safe and effective taxiing in a highly dynamic environment. We argue here for the potential for significantly reducing fuel emissions, fuel costs, and community noise, while addressing the added complexity of air terminal operations by increasing efficiency and reducing human workload. This paper describes work-in-progress for developing concepts and capabilities for autonomous engines-off taxiing using towing vehicles.

Existing ship scheduling approaches either ignore constraints on ship draft (distance between the waterline and the keel), or model these in very simple ways, such as a constant draft limit that does not change with time. However, in most ports the draft restriction changes over time due to variation in environmental conditions. More accurate consideration of draft constraints would allow more cargo to be scheduled for transport on the same set of ships. We present constraint programming (CP) and mixed integer programming (MIP) models for the problem of scheduling ships at a port with time-varying draft constraints so as to optimise cargo throughput at the port. We also investigate the effect of several variations to the CP model, including a model containing sequence variables, and a model with ordered inputs. Our model allows us to solve realistic instances of the problem to optimality in a very short time, and produces better schedules than both scheduling with constant draft, and manual scheduling approaches used in practice at ports.