arXiv subjects

Xiaochen Liu

Publications and source records attributed to Xiaochen Liu.

28 records · Page 2Linked to original sources

arXiv2021

AutoCast: Scalable Infrastructure-less Cooperative Perception for Distributed Collaborative Driving

Autonomous vehicles use 3D sensors for perception. Cooperative perception enables vehicles to share sensor readings with each other to improve safety. Prior work in cooperative perception scales poorly even with infrastructure support. AutoCast enables scalable infrastructure-less cooperative perception using direct vehicle-to-vehicle communication. It carefully determines which objects to share based on positional relationships between traffic participants, and the time evolution of their trajectories. It coordinates vehicles and optimally schedules transmissions in a distributed fashion. Extensive evaluation results under different scenarios show that, unlike competing approaches, AutoCast can avoid crashes and near-misses which occur frequently without cooperative perception, its performance scales gracefully in dense traffic scenarios providing 2-4x visibility into safety critical objects compared to existing cooperative perception schemes, its transmission schedules can be completed on the real radio testbed, and its scheduling algorithm is near-optimal with negligible computation overhead.

cs.NI↗

arXiv2021

Analytic model for feature maps in the primary visual cortex

A compact analytic model is proposed to describe the combined orientation preference (OP) and ocular dominance (OD) features of simple cells and their layout in the primary visual cortex (V1). This model consists of three parts: (i) an anisotropic Laplacian (AL) operator that represents the local neural sensitivity to the orientation of visual inputs; (ii) a receptive field (RF) operator that models the anisotropic spatial RF that projects to a given V1 cell over scales of a few tenths of a millimeter and combines with the AL operator to give an overall OP operator; and (iii) a map that describes how the parameters of these operators vary approximately periodically across V1. The parameters of the proposed model maximize the neural response at a given OP with an OP tuning curve fitted to experimental results. It is found that the anisotropy of the AL operator does not significantly affect OP selectivity, which is dominated by the RF anisotropy, consistent with Hubel and Wiesel's original conclusions that orientation tuning width of V1 simple cell is inversely related to the elongation of its RF. A simplified OP-OD map is then constructed to describe the approximately periodic OP-OD structure of V1 in a compact form. Specifically, the map is approximated by retaining its dominant spatial Fourier coefficients, which are shown to suffice to reconstruct the overall structure of the OP-OD map. This representation is a suitable form to analyze observed maps compactly and to be used in neural field theory of V1. Application to independently simulated V1 structures shows that observed irregularities in the map correspond to a spread of dominant coefficients in a circle in Fourier space.

q-bio.NC↗

arXiv2020

Efficient Pipelines for Vision-Based Context Sensing

Context awareness is an essential part of mobile and ubiquitous computing. Its goal is to unveil situational information about mobile users like locations and activities. The sensed context can enable many services like navigation, AR, and smarting shopping. Such context can be sensed in different ways including visual sensors. There is an emergence of vision sources deployed worldwide. The cameras could be installed on roadside, in-house, and on mobile platforms. This trend provides huge amount of vision data that could be used for context sensing. However, the vision data collection and analytics are still highly manual today. It is hard to deploy cameras at large scale for data collection. Organizing and labeling context from the data are also labor intensive. In recent years, advanced vision algorithms and deep neural networks are used to help analyze vision data. But this approach is limited by data quality, labeling effort, and dependency on hardware resources. In summary, there are three major challenges for today's vision-based context sensing systems: data collection and labeling at large scale, process large data volumes efficiently with limited hardware resources, and extract accurate context out of vision data. The thesis explores the design space that consists of three dimensions: sensing task, sensor types, and task locations. Our prior work explores several points in this design space. We make contributions by (1) developing efficient and scalable solutions for different points in the design space of vision-based sensing tasks; (2) achieving state-of-the-art accuracy in those applications; (3) and developing guidelines for designing such sensing systems.

cs.CV↗

arXiv2020

Weakly robust periodic solutions of one-dimensional cellular automata with random rules

We study $2$-neighbor one-dimensional cellular automata with a large number $n$ of states and randomly selected rules. We focus on the rules with weakly robust periodic solutions (WRPS). WRPS are global configurations that exhibit spatial and temporal periodicity and advance into any environment with at least a fixed strictly positive velocity. Our main result quantifies how unlikely WRPS are: the probability of existence of a WRPS within a finite range of periods is asymptotically proportional to $1/n$, provided that a divisibility condition is satisfied. Our main tools come from random graph theory and the Chen-Stein method for Poisson approximation.

math.PR↗

arXiv2020

On Localizing a Camera from a Single Image

Public cameras often have limited metadata describing their attributes. A key missing attribute is the precise location of the camera, using which it is possible to precisely pinpoint the location of events seen in the camera. In this paper, we explore the following question: under what conditions is it possible to estimate the location of a camera from a single image taken by the camera? We show that, using a judicious combination of projective geometry, neural networks, and crowd-sourced annotations from human workers, it is possible to position 95% of the images in our test data set to within 12 m. This performance is two orders of magnitude better than PoseNet, a state-of-the-art neural network that, when trained on a large corpus of images in an area, can estimate the pose of a single image. Finally, we show that the camera's inferred position and intrinsic parameters can help design a number of virtual sensors, all of which are reasonably accurate.

cs.CV↗

arXiv2020

Grab: Fast and Accurate Sensor Processing for Cashier-Free Shopping

Cashier-free shopping systems like Amazon Go improve shopping experience, but can require significant store redesign. In this paper, we propose Grab, a practical system that leverages existing infrastructure and devices to enable cashier-free shopping. Grab needs to accurately identify and track customers, and associate each shopper with items he or she retrieves from shelves. To do this, it uses a keypoint-based pose tracker as a building block for identification and tracking, develops robust feature-based face trackers, and algorithms for associating and tracking arm movements. It also uses a probabilistic framework to fuse readings from camera, weight and RFID sensors in order to accurately assess which shopper picks up which item. In experiments from a pilot deployment in a retail store, Grab can achieve over 90% precision and recall even when 40% of shopping actions are designed to confuse the system. Moreover, Grab has optimizations that help reduce investment in computing infrastructure four-fold.

cs.CV↗

arXiv2019

Maximal temporal period of a periodic solution generated by a one-dimensional cellular automaton

We study one-dimensional cellular automata evolutions with both temporal and spatial periodicity. The main objective is to investigate the longest temporal periods among all two-neighbor rules, with a fixed spatial period $σ$ and number of states $n$. When $σ= 2, 3, 4$, or $6$, and we restrict the rules to be additive, the longest period can be expressed as the exponent of the multiplicative group of an appropriate ring. We also construct non-additive rules with temporal period on the same order as the trivial upper bound $n^σ$. Experimental results, open problems, and possible extensions of our results are also discussed.

math.DS↗

arXiv2019

Periodic solutions of one-dimensional cellular automata with random rules

We study cellular automata with randomly selected rules. Our setting are two-neighbor rules with a large number $n$ of states. The main quantity we analyze is the asymptotic probability, as $n \to \infty$, that the random rule has a periodic solution with given spatial and temporal periods. We prove that this limiting probability is non-trivial when the spatial and temporal periods are confined to a finite range. The main tool we use is the Chen-Stein method for Poisson approximation. The limiting probability distribution of the smallest temporal period for a given spatial period is deduced as a corollary and relevant empirical simulations are presented.

math.PR↗

arXiv2019

One-dimensional cellular automata with random rules: longest temporal period of a periodic solution

We study one-dimensional cellular automata whose rules are chosen at random from among $r$-neighbor rules with a large number $n$ of states. Our main focus is the asymptotic behavior, as $n \to \infty$, of the longest temporal period $X_{σ,n}$ of a periodic solution with a given spatial period $σ$. We prove, when $σ\le r$, that this random variable is of order $n^{σ/2}$, in that $X_{σ,n}/n^{σ/2}$ converges to a nontrivial distribution. For the case $σ> r$, we present empirical evidence in support of the conjecture that the same result holds.

math.PR↗

arXiv2017

Secure Pick Up: Implicit Authentication When You Start Using the Smartphone

We propose Secure Pick Up (SPU), a convenient, lightweight, in-device, non-intrusive and automatic-learning system for smartphone user authentication. Operating in the background, our system implicitly observes users' phone pick-up movements, the way they bend their arms when they pick up a smartphone to interact with the device, to authenticate the users. Our SPU outperforms the state-of-the-art implicit authentication mechanisms in three main aspects: 1) SPU automatically learns the user's behavioral pattern without requiring a large amount of training data (especially those of other users) as previous methods did, making it more deployable. Towards this end, we propose a weighted multi-dimensional Dynamic Time Warping (DTW) algorithm to effectively quantify similarities between users' pick-up movements; 2) SPU does not rely on a remote server for providing further computational power, making SPU efficient and usable even without network access; and 3) our system can adaptively update a user's authentication model to accommodate user's behavioral drift over time with negligible overhead. Through extensive experiments on real world datasets, we demonstrate that SPU can achieve authentication accuracy up to 96.3% with a very low latency of 2.4 milliseconds. It reduces the number of times a user has to do explicit authentication by 32.9%, while effectively defending against various attacks.

cs.CR↗