Industry has made extensive use of biologically-inspired computing tools, including artificial neural nets, genetic algorithms, genetic programming, agent-based modeling, differential evolution, swarm computing and artificial immune systems. Sample areas of application of evolutionary tools include data mining and symbolic regression, control of dynamic systems, combinatorial optimization for 2- and 3-D layout, plant floor scheduling, vehicle routing, and single- and multi-objective optimization for product design. BEACON’s Knowledge Transfer goal is to develop effective mechanisms and pathways to facilitate intellectual exchanges among BEACON partners and industry that will support the sharing of knowledge and application of new technology.
Dr. Betty HC Cheng, Industrial Relations Manager
To facilitate technology exchange, Dr. Betty H.C. Cheng, as the BEACON Knowledge Transfer and Industrial Relations Manager, works to facilitate technology exchange between BEACON members and industrial collaborators. That is, we want two-way knowledge transfer: we want to transfer our technical capabilities to the industrial sector, but we also want industrial problems and feedback to guide our research activities. Three key types of industrial collaborations have been established. First, BEACON researchers are working on industry-funded projects. Second, industrial collaborators provide research challenge problems to BEACON researchers, where jointly developed preliminary research results may lead to larger federally-funded projects. Finally, BEACON students may participate in industry internships, thereby providing a means for the students to better understand and work on real industry problems, and an industrial organization can directly benefit from having in-house BEACON expertise applied to their problems.
Industry-Provided Challenge Problems:
Instead of working with “toy problems,” BEACON aims to work on real problems with real data provided by industry. In early stages of a relationship with a company, rather than working with proprietary data, we ask that the company provide “sanitized” data that will enable us to provide a real solution to a real problem, without requiring the company to disclose any proprietary information. Below are examples of collaborations with industrial and federal organizations involving industrial-strength, sanitized data.
Selected Industrial and Community-targeted Activities:
- Multi-Criterion Decision Making Tools for Land Use Modeling (E. Goodman, K. Deb)
- Automated analysis of software models for feature interactions (B. Cheng)
- Automated generation of software reconfigurations for autonomous systems (B. Cheng, P. McKinley)
- Dealing with uncertainty for adaptive systems (B. Cheng)
- Autonomous Vehicles (P. McKinley, C. Ofria)
- Software reuse (B. Cheng)
- Traffic Management (R. Heckendorn)
- Big Data Analytics (T. Brown)
- NASA Tool set (G. Dozier)
- Robotic Fish (X. Tan, P. McKinley, J. Boughman)
- Multi-level and Multi-timeframe Supply Chain Optimization (K. Deb, E. Goodman)
- Design Optimization Tools for Ultra-High-Definition Structures of Structures (R. Averill, A. Diaz, K. Deb, E. Goodman)
Industrial and Federal Agency Collaborators
- Axia Institute
- BAE
- Chrysler
- Continental Automotive
- DARPA
- Dow Chemical Company
- Ford Motor Company
- General Motors
- Northrup Grummon
- NASA
- Red Cedar Technology
BEACON-Related Research Tools and Resources:
-
Avida
Avida is a scientific software platform that allows a user to experiment with populations of actively evolving computer programs. Avida runs on Linux/Unix, Mac OS X, and Windows. Avida is free to use and open source, distributed under the LGPL.
-
Avida-ED
Avida-ED is an award-winning educational application developed at Michigan State University for undergraduate biology courses to help students learn about evolution and scientific method by allowing them to design and perform experiments to test hypotheses about evolutionary mechanisms using evolving digital organisms.
(eXploration Toolset for Optimization Of Launch and Space Systems)
A suite of cross-disciplinary evolutionary optimization design tools. Applications include design of space vehicles, structural components, radiators, lunar and planetary habitats, evacuation planning, and many other space applications.
A suite of tools, including NSGA II, related to non-linear, many and multi-objective optimization developed by Dr. Kalyanmoy Deb, Koenig Endowed Chair.
-
EpiFire
Project led by Lauren Meyers, UT Austin, Biology. Meyers is developing software to study epidemiology using contact networks. This work is used for research and teaching at UT Austin, Univ of Washington
-
OpenNERO:
Simulation environment for research in multiagent systems. (Igor Valerievich Karpov; Risto Miikkulainen)
-
NEAT
Software for evolving neural networks V1.2.1. (Kenneth O Stanley; Igor Valerievich Karpov; Erkin Bahceci; Risto Miikkulainen)
-
PyEC
Software for Evolutionary Annealing and a number of other evolutionary and stochastic optimization algorithms (Alan Lockett; Risto Miikkulainen)
-
UT^2
Software for the winning entry in the BotPrize 2012 competition (Igor Valerievich Karpov; Jacob Schrum; Risto Miikkulainen)
-
ESL:
Software for Egalitarian Social Learning in the robot foraging domain (Eliana Feasley; Wesley Tansey)
-
DigitalCertainty:
A BEACON-related spin-off led by Professor Risto Miikkulainen. “Create thousands of landing pages in minutes and Digital Certainty will find the highest converting option. Increasing conversions is the key to a successful digital marketing plan. Let the system do the hard work. Stay creative.” (Digital Certainty website)
Highlights of Recent Knowledge Transfer Activities:
Robotic Fish
Professor Xioabo Tan (MSU) has filed an invention disclosure on Gliding Robotic Fish in February 2013. This technology integrates key advantages of robotic fish with those of underwater gliders, and is expected to result in underwater robots with high locomotion efficiency and high maneuverability. The technology has a myriad of applications in aquatic environmental sensing, and it was highlighted at the third annual MSU Innovation Celebration in June 2013.
Tan has received two grants, one from Spartan Innovations and the other from NSF, to advance the market readiness of the technology. These projects will make the technology easier to service and customize with longer operational time. Spartan Innovations and MSU Technologies are working closely with Tan’s group on this commercialization effort. Tan also has plans to start up a spin-off company once the current phase of development is complete, around early 2015.
Videos illustrating the technology are available on YouTube: (shows field tests at Kellogg Biological Station with fellow BEACON researcher Elena Litchman and at Kalamazoo River):
Automated Analysis of Software Models to Detect Unwanted Feature Interactions
Professor Betty Cheng (MSU) is collaborating with researchers and developers at Ford Motor Company to analyze industrial-strength models to detect unwanted properties. The models are provided by Ford and have been sanitized to remove any proprietary information. This collaboration led to the publication of a full paper in the International Conference on Model Driven Engineering and Languages (MODELS). In addition, the paper was nominated for Best Paper in the Applications Track. Cheng continues to collaborate with and receive challenge problems from Ford.
Addressing of Land Use Problems Using Evolutionary Multi-Criterion Decision Making Tools
Mr. Jonas Schwaab was a visiting scholar in BEACON in 2015-16, sponsored by a Doc.Mobility Fellowship from the Swiss National Science Foundation. He collaborated with Profs. Deb and Goodman on the problem of identifying the most appropriate sites for conversion from agricultural land to urban housing usage in several Swiss communities. Conflicting objectives were minimizing the per capita costs of providing urban services (roads, utilities, etc.) to the plots of land versus minimizing the loss of agricultural productivity by favoring the poorest agricultural land for conversion. Several papers have been jointly published by Schwaab, Deb and Goodman and Schwaab’s Ph.D. supervisors in Germany (University of Bonn) and Switzerland (ETHZ). A meeting of all involved, plus another Ph.D. student, was held in Berlin in July, 2017, where it was decided to pursue further collaboration, and potential sources of additional funding were identified.
