I'm a Research Professor and Head of the Biosciences Group of the University of Michigan Transportation Research Institute. I conduct research in a variety of areas relating to anthropometry and biomechanics, including vehicle ergonomics and vehicle occupant crash protection. I'm also a Research Professor in the Center for Ergonomics in Industrial and Operations Engineering, where I am Director of the Human Motion Simulation Laboratory. The HUMOSIM Lab develops movement simulation algorithms and ergonomics analysis tools for use with commercial human modeling software.
Follow the links at the right for more information about my research and see these highlights of recent projects.
Locating body surface landmarks on whole-body scan data is usually a time-consuming manual process. Manual digitizing is the "gold standard", producing higher accuracy and precision than automated methods, but for some applications a rapid, fully automated procedure is desireable, even if accuracy is lower. I presented a brief paper at the IEA conference in Melbourne, Australia, this month examining the precision of a purely statistical method for predicting landmark and joint locations. The method is based on a statistical body shape model that incorporates landmarks and joints. The results showed that the predictions for torso joints are surprisingly precise when scan data is fitted using a rapid optimization process. The most important near-term application is the estimation of a kinematic linkage for avatars created from Kinect scan data.
The Biosciences Group is conducting a broad range of research for the U.S. Army, including a large-scale study focused on improving protection for soldiers in vehicle crash and rollover events. At the 2015 GVSETS meeting, my colleague Dr. Jingwen Hu presented an overview of outcomes from a sled-test series comparing the performance of alternative belt restraint systems in frontal impact. This is the first study to examine the influence of restraint system configuration across a range of body size, taking into account the effects of body armor and body borne gear. This work was recognized with a "best paper" award, one of only 2 out of 53 peer-reviewed papers to receive this honor.
My colleagues Dr. Daniel Park and Dr. Jangwoon Park presented some recent work at the Applied Human Factors and Ergonomics (AHFE) annual meeting in Las Vegas. Dr. J. Park presented a new method for estimating pelvis position and orientation in automobile seats. This perennially challenging problem is more difficult in individuals with high BMI, and the new methods enable adjustments to account for larger flesh margins. Dr. Daniel Park presented methods for mapping a statistical body shape model generated on one manikin mesh to another mesh. This method allows us to apply our model outcomes much more broadly. In particular, we can rapidly generate a CATIA manikin automatically from Kinect scan data.
Analyses of crash data in the field have shown that obese occupants are at higher risk in frontal crashes than occupants of normal weight. Improving protection for obese occupants requires an understanding of how these occupants interact with restraint systems. At the 2015 ESV Conference in Gothenberg, Sweden, my colleague Jingwen Hu presented a simulation study aimed at understanding how obese occupants interact with belts in frontal impact. This work is based on a new paradigm in parametric human modeling that allows rapid, accurate morphing of complex finite element models to represent individuals with a wide range of size and shape. This study validated a set of obese occupant models using post-mortem human subject data from testing conducted at the University of Virginia.
Wang et al. (2015), A Simulation Study on the Efficacy of Advanced Belt Restraints to Mitigate the Effects of Obesity for Rear-Seat Occupant Protection in Frontal Crashes Traffic Injury Prevention, 16:S75-S83, doi:10.1080/15389588.2015.1010722
Daniel Park and I have just published the first parametric body shape model for children. The model is based on laser scans of 137 children ages 3-11 in a standing posture. We used a custom template fitting approach followed by standard PCA+regression methods to create a statistical body shape model parameterized by stature, BMI, and the ratio of sitting height to stature. This model is now available online at childshape.org. As far as we know, this is the first data-based body shape model to be made available for free online. The online versions allows for downloads of a mesh surface (STL) file along with body landmark and joint locations and a set of standard anthropometric dimensions. We expect to be putting many more models online in the next year as we published more of our body shape studies. Please contact me if you have questions about using the model in your research.
The UMTRI Biosciences Group was featured in an article in Mechanical Engineering, the monthly magazine of ASME. The work of my colleague Dr. Jingwen Hu and his students on parametric human body modeling for restraint system optimization was highlighted. The article touches on a range of activities now underway in our greoup, including whole-body scanning and body shape modeling, finite-element modeling of highly detailed human anatomy, and restraint system optimization.
My collaborator Matt Parkinson, his student Brian Pagano, and I have just published the first updated assessment of U.S. child anthroometry in a generation. During the 1970s, UMTRI researchers led by Jerry Snyder conducted two large-scale studies of child body dimensions, measuring thousands of children across the country. Since that time, children in the U.S. at each age have gotten considerably heavier, with greater differences at older ages. Dr. Parkinson and I have previously published statistical methods for adjusting a detailed dataset to match a population for which only overall body dimensions, such as stature and weight, are known. We applied a similar methodology to update the detailed dimensions in the Snyder 1977 study based on recent stature and body weight data from the U.S. National Health and Nutrition Examination Survey. This update will be valuable for anyone who creates products or environments for children in the U.S. including child restraints, furniture, and clothing. Ultimately, a new, comprehensive study of U.S. child anthropometry is needed. Contact me for an article reprint.
The final report for the Seated Soldier Study conducted by UMTRI for the US Army TARDEC is now available. The report describes the methods and results from detailed measurements of 315 soldiers at three Army posts. A detailed posture analysis was conducted for both driver and squad seating conditions. Posture-prediction models based on UMTRI's Cascade methodology are presented for both environments. The models take into account the effects of body armor and body borne gear. Whole-body laser scanning was conducted to characterize body shape with and without PPE and gear. Analysis of this rich dataset will be underway for some time. Already, we are working on accommodation models for driver and squad condition, new vehicle packaging paradigms, and methods for optimizing seat design based on these findings and data. Three-dimensional body shape modeling using a subset of data from the Seated Soldier Study was used to develop anthropometric specifications for the midsize-male WIAMan ATD. If you have additional ideas on how these data could be used to improve the design of vehicles, seats, and protective equipment for soldiers, please contact me.
Building on recent work on body shape modeling and using Kinect as a body scanner, I presented a paper at the 2014 HFES conference in Chicago on work with Siemens on generating subject-specific Jack models. In May, we presented work at the 2014 DHM conference on the implementation in Jack of standing male and female statistical body shape models (SBSM). For some applications, it is useful to create a Jack manikin of a particular individual, for example, a subject in a laboratory study. Normally we would take a set of a dozen or so standard anthropometric measures and type them into Jack to scale a custom figure. That figure will have roughly the right size, but often the shape is quite unlike the individual. In the current work, we find the set of principal component scores in our male or female shape model that produces the body shape most closely matching the data from a snapshot taken with the Kinect sensor.We then pass those PC scores to Jack to obtain a figure with very similar size and shape. The paper shows quantitative comparisons for four women scanned with the Kinect system. In future work, we'll apply the new Kinect 2 sensor, which promises greater accuracy, and relax the current restrictions on scanning posture.
I presented a short communication at the 2014 IRCOBI conference in Berlin this month addressing driver knee locations. In modern vehicles, the underside of the instrument panel is designed to absorb energy in frontal impacts by exerting controlled force on the driver's knees. The knee bolster is an important component of the restraint system, sharing load with the steering wheel airbag and three-point belt. In recent work, we showed that the lap portion of the belt fits considerably more loosely for most people than for crash dummies. A loose belt means that the occupant will translate further forward before belt force builds up, potentially changing the load sharing between the belt and knee bolster. The starting knee location at the time of the crash is one critical determinant of load sharing. We used data from 100 men and women with a wide range of body size who sat in a laboratory vehicle mockup in 9 different vehicle configurations spanning the range from sports cars to SUVs. We used regression analysis to model the location of the forward-most margin of the patella (kneecap) as a function of vehicle and driver variables. As expected, the vehicle configuration had a strong effect, but we also found that taller drivers' knees are more rearward, on average. These results are useful for understanding the distribution of drivers' knee locations in any particular vehicle and could be used to conduct parametric studies of load sharing for a range of frontal impact conditions.
I presented an update of our virtual seat fit assessment work at the 8th Annual Automotive Seating Innovators Summit in Detroit. This collaboration builds on work my colleague Jingwen Hu presented at the SAE Congress in 2013. The critical insight behind this work is that rigorous dimensional assessments of seats requires evaluation with hundreds or even thousands of people. Since it's not remotely practical to do that with physical prototypes, virtual seat fit evaluations with synthesized populations of sitters is the only way to conduct an accurate dimensional analysis considering all geometric aspects of the seat sitter interaction, rather than just a few standard anthropometric dimensions. The previous work used 3D body shape models based on CAESAR. The work now underway will apply body shape models based on UMTRI data gathered in a number of studies.
The Automotive Research Center at the University of Michigan held its Annual Review this month, celebrating 20 years of research in modeling and simulation of ground vehicles. I presented an overview of some of our activities in Thrust Area II, Human-Centered Modeling in Simulation. Among other projects, we are studying the effects of body armor and body borne gear on seated reach difficulty and capability; developing new statistical tools for vehicle interior layout based on soldier posture data; and conducting sled tests and finite-element simulations to optimize belt restraints and airbags for tactical vehicles.
My colleague Daniel Park presented a paper at the 3rd International Digital Human Modeling Symposium on our work with using Kinect sensors for body scanning.
Our technique uses only two Kinect sensors and requires only about 12 seconds to obtain a subject-specific avatar. The key innovation is a rapid application of a statistical body shape model based on body scan data. The method is demonstrated using children between ages 3 and 11.
Prof. Matt Parkinson of The Pennsylvania State University presented some joint work on human modeling at the 3rd International Digital Human Modeling Symposium.
The paper reports a collaboration with Siemens to implement a statistical body shape model based on scan data in the Jack human modeling software. To our knowledge, this is the first time a widely used commercial ergonomics tool has included a high-resolution body shape model based on a statistical analysis of scan data.
This paper won an Applied Research Award at the conference.
I was fortunate to have the opportunity to present an overview of our research on tactical vehicle occupant protection at the U.S. Military Academy at West Point. LTC Bruce Floersheim, Director of the Center for Innovation and Engineering, hosted my visit.
The research I presented was funded by TARDEC through the Automotive Research Center at the University of Michigan. The ARC is a U.S. Army Center of Excellence in modeling and simulation of ground vehicles.
I gave a briefing at the U.S. Army Tank-Automotive Research, Development, and Engineering Center this month on our Seated Soldier Study. In close collaboration with the Army, and with assistance from Anthrotech, we measured the seated postures and body shapes of over 300 soldiers. This study is the first we are aware of to document in detail the effects of body armor and body-borne gear on supported seated postures. Major outcomes of this work include new posture-prediction models for drivers and squad members in military vehicles. In addition to posture measurements, over 8200 whole-body surface scans were obtained using a laser scanner, documenting male and female body shape in up to 20 postures. Along with our related work on civilian vehicle occupants, this study provides the first large-scale data on body shapes in supported seated postures. We have used the data to generate statistical models of body shape for use in a wide range of engineering applications.
The Seated Soldier Study was funded by TARDEC through the Automotive Research Center at the University of Michigan. The ARC is a U.S. Army Center of Excellence in modeling and simulation of ground vehicles.
©2015 Matthew P. Reed and The University of Michigan