David Ray Thompson

David R. Thompson


I live in Pasadena and work with the Machine Learning and Instrument Autonomy Group at the California Institute of Technology's Jet Propulsion Laboratory. My work involves computer vision, machine learning and their applications to robotic space and planetary exploration. Here are some of my research projects and personal interests. You may also be interested in my JPL webpage.

Contact: contact


I'm leading the TextureCam project to integrate automatic image analysis into a Field Programmable Gate Array. This will permit automatic image understanding by remote robotic explorers like spacecraft and planetary rovers. We aim to provide these robots the ability to identify basic science features so that they can make simple decisions about where to collect data and what to transmit to Earth. Image: automatic classification of Mars Exploration Rover image to find rocks. Original credit NASA/JPL.


Future primitive body exploration missions may need to revise trajectories and observation plans to quickly characterize the target for safe, effective observations. However, light time delays for communication with Earth may be tens of minutes to hours. When appropriate, time-critical decisions could be automated and shifted to the spacecraft for immediate access to instrument data. Mirrored planning systems on both sides of the light-time gap permit fluid transfer of authority as needed. Image: automatic detection of Hartley 2 comet plumes. Original image credit: NASA/JPL/UMD.
  • K. L. Wagstaff, D. R. Thompson, B. Bue, T. Fuchs. Autonomous Real-time Detection of Plumes and Jets from Moons and Comets. The Astrophysical Journal (in press), 2014.
  • D. R. Thompson, M. Bunte, R. Castano, S. Chien, R. Greeley. Image Processig Onboard Spacecraft for Autonomous Plume Detection. Planetary and Space Science Vol. 62, 2012. p. 153-159.(PDF)
  • D. R. Thompson, J. C. Castillo-Rogez, S. A. Chien, R. Doyle, T. Estlin, D. Mclaren. Agile Science Operations: A New Approach for Primitive Bodies Exploration. SpaceOps, 2012. Stockholm.
  • D. R. Thompson, S. Chien, D. Tran, M. Bunte, R. Greeley. Autonomous Onboard Science Data Analysis for Comet Missions. i-SAIRAS 2012. (PDF)
  • (more)


New radio astronomy instruments like the Square Kilometre Array (SKA) will generate petabyte data volumes. These investigations would benefit from real time anomaly detection and data mining to identify key features of interest. We're currently investigating adaptive resource allocation strategies and cost-sensitive computing to identify transient radio sources in radio astronomy data.
  • K. Wagstaff, N. Lanza, D. R. Thompson, T. Dietterich and M. Gilmore. Guiding Scientific Discovery with Explanations. Assoc. for the Advancement of Artificial Intelligence (AAAI) 2013.
  • C. Trott., S. J. Tingay, R. B. Wayth, D. R. Thompson, A. T. Deller, W. F. Brisken, K. L. Wagstaff, W. A. Majid, S. Burke-Spolaor, J-P R. Macquart, D. Palaniswamy. A framework for interpreting fast radio transient search experiments: application to the V-FASTR experiment. The Astrophysical Journal 2013 (in press) (preprint PDF)
  • D. R. Thompson, W. F. Brisken, A. T. Deller, W. A. Majid, S. Burke-Spolaor, S. J. Tingay, K. L. Wagstaff, and Randall B. Wayth. Real time adaptive event detection in astronomical data streams: lessons from the Very Long Baseline Array. IEEE Intelligent Systems 2013 (in press).(Preprint PDF)
  • R. Wayth, S. J. Tingay, A. Deller, W. F. Brisken, D. R. Thompson, K. Wagstaff, and W. Majid. Limits on the event rates of fast radio transients from the V-FASTR experiment. The Astrophysical Journal Letters 753:L36 2012.(PDF)
  • (more)


The Orbiting Carbon Observatory 2 (OCO-2) mission will replace the OCO spacecraft that was lost in a 2009 launch vehicle failure. OCO-2 is slated for orbit in 2014, and will provide unprecedented accuracy and coverage for measuring global atmospheric CO2 (Image: NASA/Caltech/JPL). It will give new insight into carbon sources, sinks and their sensitivity to climate change. A team of atmospheric scientists, technologists, spectroscopists is paving the way with physical models of gas absorption so that we can interpret the data when it becomes available.
  • D. R. Thompson, D. C. Benner, L. R. Brown, D. Crisp, V. M. Devi, Y. Jiang, F. Oyafuso, D. Wunch, R. Castano, C. E. Miller. Atmospheric Validation of High Accuracy CO2 Absorption Coefficients for the OCO-2 Mission, Journal of Quantitative Spectroscopy and Radiative Transfer Volume 113, Issue 17, November 2012, Pages 2265–2276.
  • D. Crisp, the OCO-2 science and algorithm teams, The ACOS CO2 retrieval algorithm Part II: Global XCO2 data characterization. Atmos. Meas. Tech., Vol. 5, 2012. 687-707.
  • (more)


Statistical models of the environment can tell explorer robots where to travel, what samples to collect and what data to return to scientists. Much of my work aims to ground autonomous science decisions by remote explorers in formal principles of information theory, active learning and experimental design.
  • D. R. Thompson, N. A. Cabrol, M. Furlong, C. Hardgrove, B. K. H. Low, J. Moersch, D. S. Wettergreen. Adaptive Sensing of Time Series with Application to Remote Exploration. International Conference on Robotics and Automation, 2013 (PDF).
  • B. K. H. Low, J. Chien, J. Dolan, S. Chien, D. R. Thompson. Decentralized Active Robotic Exploration and Mapping for Probabilistic Field Classification in Environmental Sensing. The Eleventh International Conference on Autonomous Agents and Multiagent Systems (AAMAS), 2012.
  • D. Hayden, S. Chien, D. R. Thompson, R. Castano. Using Clustering and Metric Learning to Improve Science Return of Remote Sensed Imagery. ACM Transactions on Intelligent Systems , Vol. 3, No. 3, 2011. (PDF)
  • D. R. Thompson, D. Wettergreen, and F. Calderon P. Autonomous Science for Large-Scale Robotic Survey.Journal of Field Robotics, Vol. 28, No. 4, July/Aug 2011.(PDF)
  • (more)


Much of my work involves data mining on spectral images to automatically draft maps and characterize science phenomena. Image courtesy NASA/JPL.
  • D. R. Thompson, M. de la Torre Juarez, C. M. Barker, J. Holeman, S. Lundeen, S. Mulligan, T. H. Painter, E. Podest, F. C. Seidel, E. Ustinov. Airborne imaging spectroscopy to monitor the evolution of urban mosquito microhabitats. Remote Sensing of Environment, Vol. 137, Oct. 2013. (Preprint PDF)
  • D. R. Thompson, B. Bornstein, S. Chien, S. Schafffer, D. Tran, B. Bue, R. Castano, D. Gleeson, A. Noell, Autonomous Spectral Discovery and Mapping onboard the EO-1 Spacecraft. IEEE Transactions on Geoscience and Remote Sensing , Vol. 51 No. 6, June 2013. (Preprint PDF)
  • D. R. Thompson, L. Mandrake, R. O. Green, S. Chien, A Case Study of Subpixel Target Detection in Multimodal and Outlier-contaminated Scenes IEEE Geoscience and Remote Sensing Letters. (Preprint PDF)
  • M. Gilmore, D. R. Thompson, L. J. Anderson, N. Karamzadeh, L. Mandrake, R. Castano. Superpixel segmentation for analysis of hyperspectral datasets, with application to CRISM data, M3 data and Ariadnes Chaos, Mars. Journal of Geophysical Research, Vol. 116, E07001, 2011.
  • D. R. Thompson, L. Mandrake, M. S. Gilmore and R. Castano, Superpixel Endmember Detection. Transactions on Geoscience and Remote Sensing, 48(11): 4023-4033, Nov. 2010.
  • (more)

Other Interests


The humble leaf about which the world revolves! Lots of tea-heads will wax poetic about the 300-year-old British teamaking tradition. I favor the 2000-year-old Asian tradition: an infinite variety of styles and flavors, each refined to its own art form.
  • Fujian Whites (a.k.a. Bai Mu Dan, Shou Mei): I prefer these rustic styles of white tea to their classier "Silver Needle" cousins. They look like craft-store potpourri, and smell like coconut or (sometimes) apricots. Shou Mei in particular makes a sturdy daily cuppa.
  • Sencha: As Japanese as Godzilla, but greener still, this classification encompasses a wide range of daily (non-ceremonial) Japanese teas. It looks like grass clippings, and tastes like vegetables - spinach or asparagus, perhaps. This is a tough one to brew properly, especially at the higher grades. But expertise is rewarded with a well-nuanced cup.
  • Tung-Ting Oolong: The greener oolong teas, especially those produced in Taiwan, are my favorite of all. Unfortunately their price is usually commensurate with their quality! Flavors suggest honey, peach, vanilla, or a bouquet of flowers. There's no tea presentation quite so elegant as dozens of perfect uniform two-leaves-and-a-bud pickings.


I practice a calorie restriction regimen. The idea is based on a large number of animal studies (from nematode worms up to rats and monkeys) which show that reducing calorie intake while maintaining nutrient levels triggers health-improving metabolic changes and slows the aging process by as much as 30-40%. What about humans? I'll have a better idea in 2050, but the preliminary tests (T3, WBC, lipids, body temperature, etc.) suggest encouraging parallels to the animal studies. These days I consume 1800 calories a day under the supervision of a physician.


A relatively new music craze. I'm a big fan of the French Impressionists (Ravel, Debussy) although I've recently investigated some more modern stuff. Bartok's string quartets are my all-time favorite.

Copyright 2013 David Ray Thompson, some images and papers excepted.

Valid XHTML 1.0 Transitional