InvasIC D1 - Invasive Software-Hardware Architectures for Robotics
In subproject D1, we will explore techniques of self-organization to efficiently allocate available resources for the timely varying requirements of robotic applications. We expect that less computing resources are needed to fulfill the application requirements compared to traditional resource assignment at compile-time.
To be autonomous, humanoid robots should be able to learn to operate in the real world and to interact and communicate with humans. They have to model and reflectively reason about their perceptions and actions in order to learn, act, predict and react appropriately.
Performing these tasks in the real world, especially in real-time, demands not only substantial high computing power but also concurrent hardware and software architectures that support situation-dependent context switching between different applications such as natural dialog management, visual perception of the environment, situation interpretation, task and motion planning, as well as action execution. Such architectures should allow to process the large amount and variety of sensory data in parallel and allocate resources appropriately.
In InvasIC, we will investigate the implementation of a cognitive robot control architecture with its different processing hierarchies using invasive hardware technologies, language and architectural methodologies. The goal is to explore techniques of self-organization (by means of invade/retreat) to efficiently use available resources for the timely varying requirements of robotic applications.
In particular, we will investigate a) the partitioning of robotics algorithms, e. g., computer vision, task and motion planning on both tightly-coupled processor arrays and heterogeneous RISC clusters, and b) the adaptation of such algorithms in order to efficiently use the available computing resources.
In Subproject D1 we are working in close cooperation with the KIT
Humanoids and Intelligence Systems Lab - Institut für Anthropomatik
Invasive computing for timing-predictable stream processing on MPSoCs. it - Information Technology, 2016 mehr… BibTeX Volltext ( DOI )
Resource-Aware Programming for Robotic Vision. First Workshop on Resource awareness and adaptivity in multi-core computing; co-located with IEEE European Test Symposium (ETS), 2014 mehr… BibTeX
Improving Efficiency of Embedded Multi-core Platforms with Scratchpad Memories. 1st International Workshop on Multi-Objective Many-Core Design (MOMAC) in conjunction with International Conference on Architecture of Computing Systems (ARCS), 2014 mehr… BibTeX
Resource Prediction for Humanoid Robots. First Workshop on Resource awareness and adaptivity in multi-core computing; co-located with IEEE European Test Symposium (ETS), 2014 mehr… BibTeX
Resource-Aware Harris Corner Detection based on Adaptive Pruning. Architecture of Computing Systems (ARCS), 2014 mehr… BibTeX
Potentials and Challenges for Multi-Core Processors in Robotic Applications. Workshop "Roboterkontrollarchitekturen" auf der Informatik 2013, 43. Jahrestagung der Gesellschaft für Informatik, GI-Edition "Lecture Notes in Informatics" (LNI), 2013 mehr… BibTeX
Acceleration of Optical Flow Computations on Tightly-Coupled Processor Arrays. PARS, 2013 mehr… BibTeX
A Resource-Aware Nearest Neighbor Search Algorithm for K-Dimensional Trees. Conference on Design and Architectures for Signal and Image Processing (DASIP), 2013 mehr… BibTeX
Invasive Computing for Robotic Vision. Asia South Pacific Design Automation Conference (ASP-DAC), 2012 mehr… BibTeX
Real-Time Motion Detection Based On SW/HW- Codesign for Walking Rescue Robots. In: Real Time Image Processing. Springer Journal of Real Time Image Processing, 2012 mehr… BibTeX