Beyond animation

In 1998, nearly three million late night television viewers, tuned to Germany's Stern TV, watched technicians scan the national beauty queen with a laser-based, whole-body measuring system. Seconds later, after one pass, they witnessed her three-dimensional digital twin as it appeared on computer.

The demonstration highlighted an ongoing survey known as CAESAR (Civilian American and European Surface Anthropometry Resource). Administered by the Society of Automotive Engineers (SAE), CAESAR is the first international anthropometric survey to use a whole body scanner for data collection. As a result, the project revises current human measurement databases with detailed, high-resolution measurement of the body's surface.

Automotive design engineers, at least of the male persuasion, should find these figures as fetching as Miss Germany's. Adapted into human modeling software tools like RAMSIS and JACK, anthropometric surface data promise improvements in driver positioning and safety, as well as interior layouts. Indeed, human modeling is already impacting the manner in which cars are designed.

Meet RAMSIS, Jack, and Jill

Fat, skinny, short, or tall--90 statistically based anthropometric types, derived from any database, can be applied to RAMSIS software, TechMath's CAD tool for ergonomic design. Developed in conjunction with the German automobile industry, RAMSIS permits the integration of human modeling data into CAD systems like Dassault's CATIA and SDRC's I-DEAS for simulating occupant motion associated with tasks such as foot pedal operation, gear shifting, and control panel access.

Built-in algorithms can also analyze occupant comfort, field of vision, and safety restraint routing, as well as verify adherence to pertinent global standards. Using RAMSIS, engineers can perform extensive analysis of ergonomics and design during the early stage of product development without physical mockups. Structural features associated with the software include wire frame and surface model depiction of the virtual human, kinematically correct joint motion, and even different shoe models (for pedal design and position).

Similarly, Engineering Animation Inc. helps manufacturers design vehicle interiors for optimal occupant comfort and performance with Jack OPT (Occupant Packaging Toolkit). Like RAMSIS, EAI's Jack (and Jill, for female anthropometrics) software inserts virtual humans into digital environments to evaluate the ergonomics of design before building costly prototypes. Such tools enable automotive design engineers to:

Scale humans using validated population databases

• Predict occupant postures based on user defined tasks

• Define the maximum reach capacity for any given person

• Calculate movements needed for certain tasks

• Perform real-time visual simulation

Perhaps engineers wish to see how the hand holds the steering wheel, or how it grips the shifter from a certain seating posture? What if the driver is wearing a bulky down parka? Programs within RAMSIS or Jack let the designer take such considerations into account. Both tools, moreover, run on UNIX or Windows NT.

Starting point for design

Design of a vehicle's interior layout begins with what the SAE calls the Seating Reference Point, or SgRP. Placing the occupant's assumed hip center at this point in space helps the seating design meet mandated safety regulations. Included as part of the vehicle's CAD data, the SgRP must fall within an envelope that accounts for design variables like cushion deflection and the seat's range of motion (forward/back, up/down).

The corresponding point on the human model is the H-point, or point midway between the model's hip joints. Based on the 50^th percentile person, and defined by the SAE, the hip center enables the designer to correctly position the human model within the digital vehicle.

The designer's goal, explains Ernst Assmann, a group leader in the ergonomic department at BMW, is to locate the H-point within the SgRP envelope. "A smaller model might be situated at the envelope's forward end, while a big person could be at the back end, but the H-point must correspond to the seat reference point."

To match the H-point and SgRP, the designer imports the appropriate CAD data into the Human Modeling software. In one portion of the screen, the CAD data describes the vehicle layout in 2D, 3D, wireframe, or solids. The same-scale human model, incorporating the desired anthropometrics, resides elsewhere on the screen.

With a click of the mouse, the Human Modeling software superimposes the two drawings, matching H-point and SgRP. Once the model is positioned correctly within the digital vehicle, simulation software will accurately determine ergonomic realities such as headroom, leg room, access to controls, and interference with hand brake application or other operational movements.

"For those automakers that are looking to cut weeks, or even months, off their development cycles," observes John Carlton, human factors engineer at TecMath, "products like RAMSIS are going to be essential in reaching this goal." What's next? Carlton claims human modeling systems that encompass virtual reality tools are enabling engineers to actually immerse themselves into the digital design environment.

Cadavers, surrogates, and simulation

If you have considered donating your body to science upon death, perhaps you should first picture this: An abrupt transfer of your abandoned husk from its tranquil resting place in cold storage to the bright lights of an automotive test facility. Technicians slice the body open, insert the appropriate accelerometers and load cells, and seat the corpse in a prototype vehicle--buckled up for safety, of course. Then they crash the car into a brick wall. None of this hurts--you are already dead--but it could be a bit disconcerting, even when viewed from the afterlife.

Yet vehicle safety depends on data from such activity. Captured by instrumentation and high-speed video, and supplemented by input from human volunteers and surrogates (i.e., physical dummies), crash test data helps automotive engineers predict how occupants within a vehicle will move, what parts of the vehicle they will impact, and how hard they will hit those parts upon collision.

Real crash testing, however, is expensive. That's why safety design also depends on importing mathematical models of crash dummies into software such as MADYMO (MAthematical DYnamic MOdels), developed by TNO Automotive, headquartered in Delft, The Netherlands.

Virtual dummies. MADYMO links the computerized dummies with vehicle CAD data to perform crash simulation and safety analysis directly on the computer, saving car manufacturers the time and expense of real crashes. Simulation results, in turn, help engineers refine and improve real dummy biofidelity.

Because MADYMO allows Multibody and/or Finite Element Modeling (FEM), users can create a wide variety of customized models to evaluate the effectiveness of specific restraint systems--seat belts or air bags, for example. Easy-to-generate Multibody systems can be simply described as rigid and flexible bodies interconnected by various types of kinematic joints (revolute, spherical, translational, etc.).

While current dummy models comprise dozens of parts, a simple mathematical dummy model might have six parts: head, torso, left/right arm, and left/right leg. The connecting joints accept an assigned stiffness in the form of spring/damper systems. This quick, computationally efficient method of modeling dynamic systems lends itself to large parametric design studies.

Finite Element capabilities have proven effective in modeling deformable contact. MADYMO's explicit transient finite element solver includes a choice of brick, shell, membrane, beam, and truss elements. Many material models are available including metals, rubbers, fabrics, and foams.

A big advantage of MADYMO, says TNO Technical Support Engineer Rob Marshall, is that both techniques can be blended to create the most efficient model for a specific application. For example, the whole dummy might be defined by rigid body modeling, while Finite Element Modeling details the seat belt.

"FEM is slow and requires a lot of computer time compared to Multibody modeling. If a design requires 100 tests, 20 minutes Vs 20 hours per test is a big difference. An efficient combination of multibody and FEM speeds the process."

Human modeling enhances the results of traditional crash testing by 1) giving design engineers more data to create better biofidelic dummies, and 2) providing input for simulation via MAthematical DYnamic MOdeling (MADYMO).

Virtual humans. Unfortunately, cadavers and dummies--real or virtual--don't necessarily react in a crash situation like live victims. Even human volunteers subjected to low-speed impact tests will tense up involuntarily, skewing the collected data.

In addition, because crash testing is so very expensive, the procedure has been limited to dummies and dummy models whose body sizes define the biggest range of vehicle occupants, as well as the average size occupant. Specifically, the 50^th percentile male, the 5^th percentile female, and the 95^th percentile male. This results in an incomplete set of data, as TNO's Marshall explains:

"Right now, we might determine that in a frontal impact crash a certain car performs great for the 50^th percentile male, provides adequate protection for the 5^th percentile female, but is horrible for the 95^th percentile man. Now, does that mean it is safe up until the 90^th percentile and from 90 to 95 protection drops off? Or, does it mean that the car's safety design is good to the 51^st percentile and everything between 51 and 95 is bad?"

Fortunately, mathematical models of real humans are beginning to complete the picture. Substituting RAMSIS, JACK (main article), or the TNO Human Model for mathematical models of crash-dummies, coupled with new methods of scaling, will allow simulation of crash tests using a wide variety of body types and sizes.

Where will this research lead? Marshall predicts radical changes in vehicle design as human mathematical models become ever more detailed.

'Does this suit make me look old?'

By Anna Kochan, Contributing Editor

In an attempt to develop vehicles adapted to the needs of older drivers, Ford has developed a 'third-age' suit for its design teams. Looking like a cross between a beekeeper's protective gear and an astronaut's space suit, the third-age suit is made from materials that add bulk and restrict movement in certain areas of the body, such as the knees, elbows, torso, and neck. It also incorporates gloves which reduce the sense of touch and spectacles that simulate deteriorating eyesight.

Designers on the Focus, the first Ford car to benefit from the development, wore the suit to assess the difficulty of getting in and out of a number of test vehicles, as well as the ease of accessing controls. Phil Hale, a spokesman at Ford, says this exercise resulted in significant design changes on the Focus.

"To get in and out of the Focus, you don't have to twist as much as you do on other cars because the hip swivel point has been raised," he notes. Other benefits: doors that open wider, rubber moldings over controls, and improved dashboard illumination.

Designed in collaboration with ICE Ergonomics, a wholly-owned subsidiary of UK's Loughborough University, the third-age suit is based on videos that were taken of volunteers getting in and out of cars. "These showed how older persons entered and exited a car differently than younger people," confirms Sharon Cook, principal consultant at ICE. "Using this information, together with existing scientific knowledge about the aging process, we constructed the suit."