A High-Performance System-on-Chip Architecture for Direct Tracking and SLAM

Add to your list(s) Download to your calendar using vCal

• Konstantinos Boikos (Imperial College)
• Monday 06 March 2017, 15:00-16:00
• Room 610B, EEE Building.

If you have a question about this talk, please contact George A Constantinides.

Simultaneous Localisation and Mapping, or SLAM is a family of algorithms that solve the problem of estimating an observer’s position in an unknown environment, while generating a map of that environment. SLAM algorithms that produce high quality, dense maps require powerful and heavy hardware. An efficient, low-power design that can implement such functionality in hardware can enable many exciting new applications in robotics. This work consists of a very high bandwidth streaming architecture that has at its centre a direct photometric tracking core that implements a non-linear least-squares optimisation for direct whole-image alignment. We will also discuss novel hardware architectures that can push the efficiency (performance to area and power ratio) further with SLAM in mind. An initial implementation tested on a Zynq – 7020 SoC already achieves an acceleration of approximately 20 times in comparison to a baseline system running on a dual-core ARM -A9, but the architecture can already achieve 1.5x that on a slightly bigger board with only a 20-30% increase in resources. The tracking core is also designed to be compatible with multi-pipeline designs processing the new frames in a round-robin fashion, which means we can achieve an almost linear increase in performance with extra resources as long as there is enough DRAM bandwidth.

This talk is part of the CAS Talks series.

This talk is included in these lists:

• Andrea Picciau's list
• CAS Talks
• Circuits and Systems Group: Internal Seminars
• Room 610B, EEE Building

Note that ex-directory lists are not shown.