Physical Computation Laboratory

Department of Engineering, University of Cambridge

All of our research activities build on a simple unifying idea: exploit information about the physical world to make more efficient computing systems that interact with nature.


We apply this single idea to investigate new approaches to sensors, sensor I/O, new processor architectures for sensor systems, machine learning from sensor data, and actuators/displays that exploit properties of human perception. We conduct fundamental research augmented with hardware prototypes to get the results of our research out into the world.

Prospective Ph.D. students: First apply to the graduate program before getting in touch. We usually only consider students who have previously completed a successful internship/UROP or M.Eng. project in the group. For a list of some of the available student research projects, see here.

Physical Computation Laboratory Logo

Recent News



Selected Recent Publications and Preprints



Research Highlights


Exploiting Physics in Sensing and Computation  New

We investigate how constraints imposed by physics on sensor signals can enable new ways of obtaining data from sensors while protecting privacy. We use insights about measurement uncertainty and how it propagates computations to investigate new programming languages, processor architectures, and microarchitectures for computing on physical measurements from sensors.


Related Publications
  • P. Stanley-Marbell and M. Rinard, "Warp: A Hardware Platform for Efficient Multimodal Sensing With Adaptive Approximation. IEEE Micro, vol. 40, no. 1, pp. 57-66, 1 Jan.-Feb. 2020.
  • V. Tsoutsouras, M. Vigdorchik, and P. Stanley-Marbell, “Synthesizing Compact Hardware for Accelerating Inference from Physical Signals in Sensors". Under review, 2020.
  • G. Brooks, Y. Wang, P. Stanley-Marbell, “Safeguarding Sensor Device Drivers Using Physical Constraints". Poster, EuroSys 2019, March 2019. [Poster PDF]
  • H. Sarson, R. Voo, P. Stanley-Marbell, “Evaluating RISC-V Instructions Natively with Narvie". Poster, EuroSys 2019, March 2019. [Poster PDF]
  • J. Lim and P. Stanley-Marbell. "Newton: A Language for Describing Physics". ArXiv Preprint, November 2018. [ArXiv Preprint]

  • Card image cap

    Our Warp highly-integrated multi-sensor experimental research testbed serves as the vehicle for many of our research ideas. We revise the design multiple times a year and make both the hardware and firmware open source to support other researchers.




    Exploiting Physics for Low-Power Sensing and I/O  Active

    We investigate techniques that reduce the activation and I/O energy used by sensors such as accelerometers used in everything ranging from wearable health monitors to industrial and environment monitoring sensors.


    Related Publications
  • P. Stanley-Marbell and M. Rinard, “System, method, and apparatus for reducing power dissipation of sensor data on bit-serial communication interfaces". US Patent US 10,135,471 B2, granted 20th November 2018. [Google Patents Link]
  • P. Stanley-Marbell and M. Rinard, “Method and Apparatus for Reducing Sensor Power Dissipation". US Patent App. 15/596,568, April 2018. [Google Patents Link]
  • P. Stanley-Marbell and P. Hurley, “Probabilistic Value-Deviation-Bounded Integer Codes for Approximate Communication”. Under review, 2018. [ArXiv Preprint]
  • P. Stanley-Marbell and M. Rinard, “Error-Efficient Computing Systems”. In FnT EDA, Vol. 11, No. 4, pp. 362-461, 2017. [Publisher Link]
  • P. Stanley-Marbell, P. A. Francese, and M. Rinard, "Encoder Logic for Reducing Serial I/O Power in Sensors and Sensor Hubs". In 28th IEEE Symposium on High-Performance Chips (Hot Chips'16). [Preprint]
  • P. Stanley-Marbell and M. Rinard, "Reducing Serial I/O Power in Error-Tolerant Applications by Efficient Lossy Encoding". In 53rd ACM/IEEE Design Automation Conference (DAC'16). [Publisher Link]
  • P. Stanley-Marbell and M. Rinard, “Efficiency Limits for Value-Deviation-Bounded Approximate Communication”. In IEEE Embedded Systems Letters Journal, 7(4), 109-112, 2015. [Publisher Link]

  • Card image cap

    3-bit words incurring up to 2 bit-level upsets (green).

    Card image cap

    The integers 64, 63, and their serial intra-word transitions.




    Energy-Efficient OLED Displays by Exploiting Human Color Perception  Active

    We investigate techniques that reduce OLED display power use by exploiting properties of human color vision perception.


    Related Publications
  • J. Cambronero, P. Stanley-Marbell, and M. Rinard, “Incremental Color Quantization for Color-Vision-Deficient Observers Using Mobile Gaming Data”. Under review, 2018. [ArXiv Preprint]
  • P. Stanley-Marbell and M. Rinard, “Perceived-Color Approximation Transforms for Programs that Draw”. In IEEE Micro, vol. 38, no. 4, pp. 20-29, July/August 2018. [Publisher Link]
  • P. Stanley-Marbell, V. Estellers, and M. Rinard, "Crayon: Saving Power through Shape and Color Approximation on Next-Generation Displays". In 11th Annual European Conference on Computer Systems, (EuroSys'16), April 2016. [Publisher Link]

  • Card image cap

    Power dissipation (radius) versus hue (angle) for one OLED display.




    Applying Low-Power Sensing to Additive Manufacturing  New

    Our research objective is to use in situ sensing to investigate the relationship between sensed phenomena such as in-chamber and in-powder temperature, humidity, etc., and the process of selective laser sintering (SLS). We are building a sensor-enhanced SLS printer which will provide, for each part produced, a detailed volumetric (per-build-layer of its construction) dataset of the variations in composition and information on the in-powder, in-chamber, and ambient conditions.

    Card image cap

    Pictures from our SLS printer assembly along with pictures of the inside of the chamber during a print and after printing.






    Funding




    funding list 2018