This paper deals simultaneously with the trajectory estimation and map reconstruction by means of a stereo-calibrated vision system evolving in large-scale unknown environment. problem is widely known as Visual SLAM. Our proposal optimizes execution time VSLAM framework while preserving its localization accuracy. The contributions this are structured follows. First, novel approach based on "Weighted Mean" multiple neighbor poses detailed denoted HOOFR provides estimate after computing camera...

The present study focuses on the development of an embedded smart camera network dedicated to track and count people in public spaces. In network, each node is capable sensing, tracking counting while communicating with adjacent nodes network. Each typically uses a 3D-sensing positioned downward-view but designed framework can accept other configurations. We estimation method for relative position orientation depth cameras. This system performs background modeling during calibration process,...

This paper describes a multi-FPGA architecture dedicated for the real-time imaging using Total Focusing Method (TFM) and an advanced acquisition technique called Full Matrix Capture (FMC). The maximum operating frequency is 147.3 MHz control blocks 161.3 blocks. able to perform image reconstruction at frame rate of 73 fps resolution 128 × pixels, which adequate as performance system with good characterization defects.

Simultaneous localization and mapping (SLAM) is widely used in many robotic applications autonomous navigation. This paper presents a study of FastSLAM2.0 computational complexity based on monocular vision system. The algorithm intended to operate with particles large-scale environment. was partitioned into functional blocks allowing hardware software matching CPU-GPGPU-based SoC architecture. Performances terms processing time accuracy were evaluated using real indoor dataset. Results...

The graph-based SLAM (Simultaneous Localization and Mapping) method uses a graph to represent solve the problem. allows building map of an unknown environment simultaneously localizing robot on this map. This paper presents temporal analysis 3D method. We also propose efficient implementation, OMAP embedded architecture, which is widely used open multimedia applications platform. provide optimized data structure memory access management nonlinear least squares problem related algorithm....

A monocular SLAM system uses a single-camera trying to solve the problem of simultaneous localization and mapping. The FastSLAM2.0 employs Rao-Blackwellized particle filter estimate robot pose based on set hypotheses that represent different possible trajectories, while mapping large number landmarks. most common related with such is initialization camera bearing only sensor can not provide depth observed feature. lot methods was developed for an efficient estimation unified inverse...

This article deals with the computational complexity issue of graphbased simultaneous localization and mapping (SLAM). SLAM allows a robot that is navigating in an unknown environment to build map this while simultaneously determining pose on map. Graph-based smoothing method uses graph represent solve problem. We first propose construction takes advantage incremental sparse characteristics graph-based SLAM. exploited perform several algorithmic optimizations. Second, we present study using...

To decrease the computational complexity of computer vision algorithms, one solutions is to achieve some low-level image processing on sensor focal plan. It becomes a smart device called retina. This concept makes systems more compact. increases performances thanks reduction data flow exchanges with external circuits. paper presents comparison relating two different system architectures. The first implements logarithmic complimentary metal-oxide-semiconductor (CMOS)/active pixel interfaced...

We have implemented customized SIMD 16-bit floating point instructions on a NIOS II processor. On several image processing and media benchmarks for which the accuracy dynamic range of this format is sufficient, speed-up ranging from 1.5 to more than 2 obtained versus integer implementation. The hardware overhead remains limited compatible with capacities today's FPGAs.

Unmanned Aircraft Systems (UAS) know a tremendous interest in many fields thanks to their wide range of applications. UAS are complex "system systems" because, addition stringent safety and embedded constraints, they incorporate heterogeneous interacting sub-systems. Adequate development approaches tools have be provided deal with this complexity. A literature review concerning perspective shows lack because current simulation focus on limited or isolated application. We propose system-level...

Human errors are the primary cause of powered two-wheeler crashes worldwide due to demanding control required and often ineffective rider-training programs. Literature on rider behaviour is limited, partly lack standard investigation methodologies. This work investigated differences in riding style capability a diverse set riders. It explored impact familiarisation instruction through objective metrics. Correlation with experience was particular focus. Seven riders various levels performed...

The Cell processor is a typical example of heterogeneous multiprocessor on-chip architecture that uses several levels parallelism to deliver high performance. Reducing the gap between peak performance and effective challenge for software tool developers application developers. Image processing media applications are "main stream" applications. We use Harris algorithm detection Points Interest in an image as benchmark compare parallel schemes on processor. impact DMA controlled data transfers...

We describe the design and implementation of a Guidance, Navigation Control (GNC) System for fixed wing Unmanned aerial vehicles (UAVs). This system allows an autonomous navigation by following desired trajectory defined profile airspeed, altitude list waypoints. The GNC has hierarchical architecture integrates Controller Area Network data bus, which facilitate process testing. results Hardware In Loop simulations real-time performances demonstrate soundness implemented system. article gives...

Motorcycle drivers are considered among the most vulnerable road users, as attested by number of crashes increasing every year. The significant part fatalities relates to “single vehicle” loss control in bends. During this investigation, a system based on an instrumented multi-sensor platform and algorithmic study was developed accurately reconstruct motorcycle trajectories achieved when negotiating This is used French Gendarmerie order objectively evaluate examine way riders take their...

Feature matching plays an important role in many computer vision applications, such as object recognition, scene reconstruction or image mosaicing. In this paper, we propose algorithm called Hessian ORB - Overlapped FREAK (HOOFR) which is based on the combination of detector and bio-inspired descriptor. We address some modifications related to detection description processes order enhance HOOFR reliability, speed memory fingerprint. The experiments a widely used dataset demonstrate...

One of the solutions to reduce computational complexity image processing is perform some low-level computations on sensor focal plane. This paper presents a vision system based smart sensor. PARISl (programmable analog retina-like sensor) first prototype used evaluate architecture an on-chip such and digital processor. The integrates computing units. makes more compact increases performances reducing data flow exchanges with A has been implemented as proof-of-concept. enabled us performance...