New vehicles from the automotive industry include a growing number of progressively more sophisticat

New vehicles from the automotive industry include a growing number of progressively more sophisticated embedded systems, including a variety of advanced driver assistance systems (ADASs), such as lane departure warning hotels in rome in trevi fountain area 1 and adaptive cruise control. For reasons of cost and technology, these innovations draw upon limited resources, leading to highly specific, optimized architectures for each application and platform. The goal is to develop dependable products capable of real-time environment recognition and decision-making for passenger safety.
The level of intelligence demanded by such systems can only be achieved by adapting high-level cognitive architectures and developing new methods for designing integrated circuits. Algorithm hotels in rome in trevi fountain area parallelism, for which hardware is especially well-suited, and the computational power of field-programmable gate arrays (FPGAs) must be exploited.
With these parameters in mind, we are exploring the possibility of scaling down artificial hotels in rome in trevi fountain area intelligence techniques (neural networks) and BB1-inspired blackboard architecture 2,3 to effectively use limited resources. Our initial focus is to detect traffic signals. hotels in rome in trevi fountain area Figure 1. Phases hotels in rome in trevi fountain area (1-4) and agents (A-C) of an execution cycle in blackboard architecture. Adaptation of the BB1 blackboard architecture 2,3 . HMI: human machine interface.
FPGAs allow implementation of intrinsically parallel techniques, such as artificial neural networks. 4,5 However, low-cost FPGAs, such as might be used in industrial applications, provide extremely limited resources. We therefore propose a new approach that stresses conceptual hotels in rome in trevi fountain area and hardware architectural design. We use a low-cost hotels in rome in trevi fountain area Altera Cyclone FPGA.
Our system is based on images from a charge-coupled device (CCD) camera installed on the dashboard hotels in rome in trevi fountain area facing the road. Images are analyzed for features that may indicate the presence of a traffic signal, such as arcs and corners. Three neural networks perform this analysis, each specialized in identifying one type of feature. They carry out phase 1 (see Figure 1 ) with execution coordinated by agent A, the manager. The position and number hotels in rome in trevi fountain area of features is analyzed in phase 2 and assigned a priority level by agent B, the rater. Results are conveyed to the driver through the `executor agent, a human-machine interface, labelled hotels in rome in trevi fountain area C.
The order in which the agents hotels in rome in trevi fountain area operate and the data and resources used in the process correspond from a conceptual viewpoint to blackboard hotels in rome in trevi fountain area architecture, 2,3 a well-known type of cognitive architecture developed to facilitate data-sharing in artificial intelligence.
Insufficient resources prevent strict implementation of blackboard architecture with our hardware. A single neural network agent requires virtually all the FPGA's resources. We use at least three agents, and so must either employ several FPGAs or share a single one. Taking the latter alternative, we have built a single hardware hotels in rome in trevi fountain area neural network (double-layer perceptron) into the FPGA. We call it the Generic Neural hotels in rome in trevi fountain area Network (GNN), and it is shared between all agents. This represents a form of hardware reconfiguration. 6
The GNN is configured for each agent in runtime. It is loaded with network weights and agent biases that are stored in memory after being previously calculated off-line with Matlab. Weights of unused neurons are set to zero. Reconfiguration and execution times are sufficiently short to allow new frame processing from the camera every 40ms, the phase alternating line (PAL) rate.
Each layer in the GNN includes its own control logic. A layer of n neurons can perform the corresponding n operations in parallel due to independent access to the memories of each neuron. To prevent hotels in rome in trevi fountain area precision hotels in rome in trevi fountain area losses, over- and underflows, the data path has been dimensioned.
Designing intelligent embedded systems for the automotive industry requires adapting high-level techniques to limited resources. By sharing the same hardware for the entire hotels in rome in trevi fountain area system, we reduced components to create a generic, runtime configurable hardware neural network. System parts and operation employ a blackboard architecture, which provides a high degree of modularity, necessary for consistent scalability. System functionality can be enhanced by adding new data (weights and biases) to memory. In the near term, the system will include pedestrian detection. The same architecture could also be easily deployed on several FPGAs, or over hybrid FPGA/digital-signal-processing/analog boards.
The functional and platform hotels in rome in trevi fountain area scalability we describe will enable future applications with increased intelligence. Many applications require advanced perception algorithms hotels in rome in trevi fountain area and sensory fusion as well as distributed sensing and action, for which blackboard hotels in rome in trevi fountain area architecture is particularly well suited. Ignacio L pez Autonomous Systems Laboratory, ASLab, Universidad Polit cnica de Madrid (UPM) Madrid, Spain Rub n Salvador, Jaime Alarc n, F lix Moreno Centro de Electr hotels in rome in trevi fountain area nica Industrial, CEI, Universidad Polit cnica de Madrid Madrid, Spain
Ignacio L pez has devoted his career to systems engineering and cognitive architectures applied to the automotive industry, including methodologies for calculating pollutant emission hotels in rome in trevi fountain area inventories and the design and development of embedded systems.
hotels in rome in trevi fountain area Rub n Salvador, with a MSc in electronic engineering from Universidad de Alcal (2004) and BSc in telecommunication engineering (UPM, 2001), is currently finishing a master's degree in industrial electronics as part of his PhD in electronic engineering at UPM. His research interests include conceiving and designing high-performance reconfigurable and adaptable systems, hardware embedded hotels in rome in trevi fountain area cognitive architectures, and digital signal processing systems.
Jaime Alarc n Celis received his bachelor's degree in electrical and mechanical hotels in rome in trevi fountain area engineering and a master's degree in engineering from UPM, where he is currently completing his PhD. He also holds a master's degree in computer science hotels in rome in trevi fountain area from the Instituto Tecnol gico y de Estudios hotels in rome in trevi fountain area Superiores de Monterrey. He served as professor and head of the electronics and control department of Tecnol gico de Monterrey (Toluca Campus). His research interests hotels in rome in trevi fountain area are neural networks, parallelism, and real-time systems.
F lix Moreno is assistant professor of electronics engineering at the ETSII-UPM. He received his PhD in telecommunication engineering from UPM. He is coordinator of the Advanced hotels in rome in trevi fountain area Driver Assistance System for Urban Project Environments (ASISTENTUR) TRA2004-07441-C03-03/AUT, with the support of the Spanish Ministry of Science, under the National R D Plan, through which the work presented here has been developed. He has extensive experience in embedded system design for the automotive and aeronautical sectors. References: 1. Jaime Alarc n, Rub n Salvador, F lix Moreno, Ignacio L pez, A New Real-Time Hardware Architecture for Road Line Tracking Using a Particle Filter , Proceedings of the 32nd Annual Conference of the IEEE Industrial Electronics Society , IECON '06, pp. 736-741, 2006. 2. Michael Wolverton, Lee Brownston, BB1 v3.2 Manual , no. KSL 94-XXDept. of Computer Science, Stanford University, 1994. 3. B. Hayes-Roth, K. Pfleger, P. Lalanda, P. Morignot, M. Balabanovic, A Domain-Specific Software Architecture for Adaptive Intelligent Systems , IEEE Transactions on Software Engineering 21 (4), pp. 288-301, 1995. 4. Y. Taright, M. Hubin, FPGA implementation of a multilayer perceptron neural network using VHDL , Proceedings of the Fourth International Conference on Signal Processing. ICSP '98 2, pp. 1311-1314, 1998. 5. J. J. Blake, L. P. Maguire, T. M. McGinnity, L. J. McDaid, Using Xilinx FPGAs to implement hotels in rome in trevi fountain area neural networks and fuzzy systems , IEE Colloquium on Neural and Fuzzy Systems: Design, Hardware and Applications (Digest No: 1997/133), May 1997. 6. Yana E. Krasteva, Eduardo de la Torre, Teresa hotels in rome in trevi fountain area Riesgo, Partial Reconfiguration for Core Reallocation and Flexible Communications hotels in rome in trevi fountain area , Proceedings of Reconfigurable Communication-centric SoC (ReCoSoC) , July 2006.