ABSTRACT The use of underground freight transportation (UFT) is gaining attention because of its ability to quickly move freight to locations in urban areas while reducing road traffic and the need for delivery drivers. Since packages are transported through the tunnels by electric motors, the use of tunnels is also environmentally friendly. We examine the use of tunnels to transport individual orders, motivated by the last‐mile delivery of goods from e‐commerce providers. The use of UFT for last‐mile delivery requires more complex network planning than for direct lines that have previously been considered for networks connecting large cities. We introduce a new network design problem based on this delivery model and transform the problem into a fixed‐charge multicommodity flow problem with additional constraints. We show that this problem, the ‐UFT, is NP‐hard, and provide an exact solution method for solving large‐scale instances. Our Benders decomposition‐based solution approach exploits the combinatorial sub‐structures of the problem in a cutting planes fashion. We provide computational results for real urban environments to build a set of insights into the structure of such networks and evaluate the benefits of such systems. We estimate the costs for implementing UFT systems and break them down into a per‐package cost. Our estimates indicate at least a 40% cost savings from using a UFT over traditional delivery models. This indicates that UFT systems for last‐mile delivery are a promising area for future research.

Hybrid wheeled-legged quadrupeds have the potential to navigate challenging terrain with agility and speed and over long distances. However, obstacles can impede their progress by requiring the robots to either slow down to step over obstacles or modify their path to circumvent the obstacles. We propose a motion optimization framework for quadruped robots that incorporates non-steerable wheels and dynamic jumps, enabling them to perform hybrid wheeled-legged locomotion while overcoming obstacles without slowing down. Our approach involves a model predictive controller that uses a time-varying rigid body dynamics model of the robot, including legs and wheels, to track dynamic motions such as jumping. We also introduce a method for driving with minimal leg swings to reduce energy consumption by sparing the effort involved in lifting the wheels. Our method was tested successfully on the wheeled Mini Cheetah and the Unitree AlienGo robots. Further videos and results are available at https://www.ais.uni-bonn.de/∼hosseini/iros2023

This paper addresses two deficiencies of models in the area of matching-based market design. The first arises from the recent realization that the most prominent solution that uses cardinal utilities, namely the Hylland-Zeckhauser (HZ) mechanism, is intractable; computation of even an approximate equilibrium is PPAD-complete. The second is the extreme paucity of models that use cardinal utilities. Our paper addresses both these issues by proposing Nash-bargaining-based matching market models. Since the Nash bargaining solution is captured by a convex program, efficiency follows. In addition, it possesses several desirable game-theoretic properties. Our approach yields a rich collection of models: for one-sided as well as two-sided markets, for Fisher as well as Arrow-Debreu settings, and for a wide range of utility functions, all the way from linear to Leontief. We give very fast implementations for these models using Frank-Wolfe and Cutting Plane algorithms. These help solve large instances with several thousand agents and goods in a matter of minutes on a PC, even for a one-sided matching market under piecewise-linear concave utility functions and a two-sided matching market under linear utility functions. In contrast, using HZ, going beyond even $n = 10$ is prohibitive. Several new ideas were needed, beyond the standard methods, to obtain these implementations. In particular, we present several lower bounding schemes, which not only help improve the convergence of our solution methods but also shed light on fairness properties of the Nash-bargaining-based models.

Widely-used Rate Control (RC) algorithms, such as those in the H.264 encoder, have certain shortcomings for time-sensitive applications such as High Definition Video Conferencing (HDVC): they either respond too slowly to available bandwidth variations, causing degradation in the perceived quality of the video session, or do not optimize video quality for a given available bandwidth. To overcome these shortcomings, we propose Dynamic Rate Control (DRC) which: 1- can adjust the bitrate of the video within a fast 4 frames or so 2- is conservative and does not waste bandwidth by unnecessarily increasing the video quality, instead saving the bandwidth as bursts for future frames, and 3- uses a moving window to limit the effect of past bursts on current bitrate. We implemented DRC in the x264 codec and used it in an actual video conferencing product from Magor Corp. The results showed that, compared to the widely-used ABR and CRF rate controllers, DRC provides better video quality and user experience, while adjusting the video bitrate faster.

Current rate control algorithms developed in the H.264 encoder are unable to address High Definition video streaming requirements over best effort networks such as the Internet. We propose a rate controller (RC) for H.264 high definition video conferencing (HDVC), specifically developed in the x264 codec: an open source and high performance H.264/AVC encoder. Called the Dynamic Constant Rate Factor (DCRF), our RC tries to provide the best possible quality in terms of the current bandwidth available at a given instance of video streaming. The proposed method is analyzed from different aspects including network bandwidth, frame size, PSNR and SSIM measurements. In addition to the objective tests, DCRF is evaluated by human observers for subjective performance evaluation. The results prove that DCRF provides better results than current RC methods for HDVC.

Videoconferencing in the context of 3D virtual environments promises better spatial consistency and mutual awareness for its participants. However, in the absence of IP Multicast and limited upload bandwidth of today's DSL connections, the feasibility of such systems in supporting even a small group of users is in question. This paper presents the design and implementation of an awareness driven 3D videoconferencing application that runs on a peer-to-peer architecture and our own End System Multicast protocol. The paper highlights the unique requirements of multiparty videoconferencing applications and presents a solution that can support 4-10 bandwidth-limited users without the need for IP Multicast capability.

The growing complexity of Virtual Reality simulations and the inclusion of a greater range and number of events and interactions in such simulations precipitates the need for the recording of these events to allow their subsequent viewing, discussion and perhaps modification. Furthermore, storage of such recordings in a format defined by an international standard allows their easy use and accessibility as well as distribution. This paper presents a framework for MPEG-4 based recording and replay of Virtual Reality sessions.

As the cost of training individuals to learn how to use new technologies increases, the need for virtual environments that facilitate training becomes more noticeable. More specifically, the use of Collaborative Virtual Environments for training in industrial applications is of special interest. What follows is a discussion of two prototypes designed and implemented for the purpose of using Collaborative Virtual Environments in training scenarios.