Search

By interpreting a traffic scene as a graph of interacting vehicles, we gain a flexible abstract representation which allows us to apply Graph Neural Network (GNN) models for traffic prediction. These naturally take interaction between traffic participants into account while being computationally efficient and providing large model capacity. We evaluate two state-of-the art GNN architectures and introduce several adaptations for our specific scenario. We show that prediction error in scenarios with much interaction decreases by 30% compared to a model that does not take interactions into account. This suggests that interaction is important, and shows that we can model it using graphs. This makes GNNs a worthwhile addition to traffic prediction systems., Comment: To be published at IEEE Intelligent Vehicles Symposium 2019

We consider adversarial examples for image classification in the black-box decision-based setting. Here, an attacker cannot access confidence scores, but only the final label. Most attacks for this scenario are either unreliable or inefficient. Focusing on the latter, we show that a specific class of attacks, Boundary Attacks, can be reinterpreted as a biased sampling framework that gains efficiency from domain knowledge. We identify three such biases, image frequency, regional masks and surrogate gradients, and evaluate their performance against an ImageNet classifier. We show that the combination of these biases outperforms the state of the art by a wide margin. We also showcase an efficient way to attack the Google Cloud Vision API, where we craft convincing perturbations with just a few hundred queries. Finally, the methods we propose have also been found to work very well against strong defenses: Our targeted attack won second place in the NeurIPS 2018 Adversarial Vision Challenge., Comment: For source code and videos, see https://github.com/ttbrunner/biased_boundary_attack

Ophthalmic microsurgery is known to be a challenging operation, which requires very precise and dexterous manipulation. Image guided robot-assisted surgery (RAS) is a promising solution that brings significant improvements in outcomes and reduces the physical limitations of human surgeons. However, this technology must be further developed before it can be routinely used in clinics. One of the problems is the lack of proper calibration between the robotic manipulator and appropriate imaging device. In this work, we developed a flexible framework for hand-eye calibration of an ophthalmic robot with a microscope-integrated Optical Coherence Tomography (MIOCT) without any markers. The proposed method consists of three main steps: a) we estimate the OCT calibration parameters; b) with micro-scale displacements controlled by the robot, we detect and segment the needle tip in 3D-OCT volume; c) we find the transformation between the coordinate system of the OCT camera and the coordinate system of the robot. We verified the capability of our framework in ex-vivo pig eye experiments and compared the results with a reference method (marker-based). In all experiments, our method showed a small difference from the marker based method, with a mean calibration error of 9.2 $\mu$m and 7.0 $\mu$m, respectively. Additionally, the noise test shows the robustness of the proposed method., Comment: *The first two authors contributed equally to this paper. Accepted by IEEE Robotics and Automation Letters (RA-L), 2018

Due to decelerating gains in single-core CPU performance, computationally expensive simulations are increasingly executed on highly parallel hardware platforms. Agent-based simulations, where simulated entities act with a certain degree of autonomy, frequently provide ample opportunities for parallelisation. Thus, a vast variety of approaches proposed in the literature demonstrated considerable performance gains using hardware platforms such as many-core CPUs and GPUs, merged CPU-GPU chips as well as FPGAs. Typically, a combination of techniques is required to achieve high performance for a given simulation model, putting substantial burden on modellers. To the best of our knowledge, no systematic overview of techniques for agent-based simulations on hardware accelerators has been given in the literature. To close this gap, we provide an overview and categorisation of the literature according to the applied techniques. Since at the current state of research, challenges such as the partitioning of a model for execution on heterogeneous hardware are still a largely manual process, we sketch directions for future research towards automating the hardware mapping and execution. This survey targets modellers seeking an overview of suitable hardware platforms and execution techniques for a specific simulation model, as well as methodology researchers interested in potential research gaps requiring further exploration., Comment: Submitted for review to ACM Computing Surveys on 24/05/2018

The rapid growth of IoT era is shaping the future of mobile services. Advanced communication technology enables a heterogeneous connectivity where mobile devices broadcast information to everything. Mobile applications such as robotics and vehicles connecting to cloud and surroundings transfer the short-range on-board sensor perception system to long-range mobile-sensing perception system. However, the mobile sensing perception brings new challenges for how to efficiently analyze and intelligently interpret the deluge of IoT data in mission- critical services. In this article, we model the challenges as latency, packet loss and measurement noise which severely deteriorate the reliability and quality of IoT data. We integrate the artificial intelligence into IoT to tackle these challenges. We propose a novel architecture that leverages recurrent neural networks (RNN) and Kalman filtering to anticipate motions and interac- tions between objects. The basic idea is to learn environment dynamics by recurrent networks. To improve the robustness of IoT communication, we use the idea of Kalman filtering and deploy a prediction and correction step. In this way, the architecture learns to develop a biased belief between prediction and measurement in the different situation. We demonstrate our approach with synthetic and real-world datasets with noise that mimics the challenges of IoT communications. Our method brings a new level of IoT intelligence. It is also lightweight compared to other state-of-the-art convolutional recurrent architecture and is ideally suitable for the resource-limited mobile applications., Comment: 7 pages, 6 figures, 1 table

The introduction of autonomous vehicles (AVs) will have far-reaching effects on road traffic in cities and on highways.The implementation of automated highway system (AHS), possibly with a dedicated lane only for AVs, is believed to be a requirement to maximise the benefit from the advantages of AVs. We study the ramifications of an increasing percentage of AVs on the traffic system with and without the introduction of a dedicated AV lane on highways. We conduct an analytical evaluation of a simplified scenario and a macroscopic simulation of the city of Singapore under user equilibrium conditions with a realistic traffic demand. We present findings regarding average travel time, fuel consumption, throughput and road usage. Instead of only considering the highways, we also focus on the effects on the remaining road network. Our results show a reduction of average travel time and fuel consumption as a result of increasing the portion of AVs in the system. We show that the introduction of an AV lane is not beneficial in terms of average commute time. Examining the effects of the AV population only, however, the AV lane provides a considerable reduction of travel time (approx. 25%) at the price of delaying conventional vehicles (approx. 7%). Furthermore a notable shift of travel demand away from the highways towards major and small roads is noticed in early stages of AV penetration of the system. Finally, our findings show that after a certain threshold percentage of AVs the differences between AV and no AV lane scenarios become negligible., Comment: 12 pages, 7 figures

Anthropomimetic robots are robots that sense, behave, interact and feel like humans. By this definition, anthropomimetic robots require human-like physical hardware and actuation, but also brain-like control and sensing. The most self-evident realization to meet those requirements would be a human-like musculoskeletal robot with a brain-like neural controller. While both musculoskeletal robotic hardware and neural control software have existed for decades, a scalable approach that could be used to build and control an anthropomimetic human-scale robot has not been demonstrated yet. Combining Myorobotics, a framework for musculoskeletal robot development, with SpiNNaker, a neuromorphic computing platform, we present the proof-of-principle of a system that can scale to dozens of neurally-controlled, physically compliant joints. At its core, it implements a closed-loop cerebellar model which provides real-time low-level neural control at minimal power consumption and maximal extensibility: higher-order (e.g., cortical) neural networks and neuromorphic sensors like silicon-retinae or -cochleae can naturally be incorporated., Comment: Accepted at IEEE Robotics and Automation Magazine on 2015-12-31

In vivo, neurons establish functional connections and preserve information along their synaptic pathways from one information processing stage to the next in a very efficient manner. Paired spiking (PS) enhancement plays a key role by acting as a temporal filter that deletes less informative spikes. We analyzed the spontaneous neural activity evolution in a hippocampal and a cortical network over several weeks exploring whether the same PS coding mechanism appears in neuronal cultures as well. We show that self-organized neural in vitro networks not only develop characteristic bursting activity, but feature robust in vivo-like PS activity. PS activity formed spatiotemporal patterns that started at early days in vitro (DIVs) and lasted until the end of the recording sessions. Initially random-like and sparse PS patterns became robust after three weeks in vitro (WIVs). They were characterized by a high number of occurrences and short inter-paired spike intervals (IPSIs). Spatially, the degree of complexity increased by recruiting new neighboring sites in PS as a culture matured. Moreover, PS activity participated in establishing functional connectivity between different sites within the developing network. Employing transfer entropy (TE) as an information transfer measure, we show that PS activity is robustly involved in establishing effective connectivities. Spiking activity at both individual sites and network level robustly followed each PS within a short time interval. PS may thus be considered a spiking predictor. These findings suggest that PS activity is preserved in spontaneously active in vitro networks as part of a robust coding mechanism as encountered in vivo. We suggest that, presumably in lack of any external sensory stimuli, PS may act as an internal surrogate stimulus to drive neural activity at different developmental stages., Comment: 23 pages, 9 figures; 2 supplementary pages, 2 supplementary figures