Posturing the Jack Figure

Jack uses an inverse kinematics algorithm to solve several types of constraints that can be placed on articulated figures. These constraints generally require several parameters to be specified: a type (e.g. point-to-point, point to plane), an end effector site, the joint chain that will be involved, and a weight, among others. We control the figure through the behavior functions associated with human figures in Jack. The various human behaviors operate by managing a number of constraints that are placed on the figure.

The system must first be calibrated to account for the operator's size. This can be done in two ways - the sensor data can be offset to match the model's size, or the model can be scaled to match the operator. Either approach may be taken, depending on the requirements of the particular situation being simulated. For motion recording, it is usually preferable to have the simulated figure scaled to match the operator. An ergonomic study of a vehicle, however, would require an arbitrary figure size; in this case, the sensor data would be offset as required.

Each frame of the simulation requires the following steps:

• Step 1: The pelvis segment is moved as the first step of the simulation. The absolute position/orientation of this segment is given by the waist sensor after adding the appropriate offsets. The figure is rooted through the pelvis, so this sensor determines the overall location of the figure.

• Step 2: The spine is now adjusted, using the location of the waist sensor and pelvis as its base. The Jack human figure includes an articulated spine that models the human spine as a chain of 18 dependent joints. These joints represent the 12 thoracic and 5 lumbar vertebrae. Four parameters are required to manipulate the spine: the initiator joint, resistor joint, resistance, and spine target position [6]. In our simulation the first three parameters are fixed, and the spine target position is extracted from the relationship between the waist and head sensors. The waist sensor gives us the absolute position of the pelvis and base of the spine, while the rest of the upper torso is placed algorithmically by the model.

    
  Figure 2: Extracting the Spine Target Vector
  

  The spine target position is a 3 vector that can be thought of as the sum of the three types of bending the spine undergoes - front/back bending or flexion, twisting or axial bending, and side or lateral bending.

  The waist sensor gives us orientation information for the base of the spine; information on the top of the spine is extracted from the head sensor via a heuristic. The target vector is computed from the differences between these two orientations. Lateral bending is found from the difference in orientation along the z axis, axial twisting is found from the difference in y orientation, and flexion is determined from the difference in x orientation (Fig. 2). Note that the ``front'' vectors in this figure indicate the front of the human. The target vector is sent directly to the model to simulate the approximate bending of the operator's spine.

• Step 3: Now that the torso has been positioned, the arms can be set. Each arm of the figure is controlled by a sensor placed on the operator's palm. This sensor is used directly as the goal of a position and orientation constraint. The end effector of this constraint is a site on the palm that matches the placement of the sensor, and the joint chain involved is the wrist, elbow, and shoulder joint.

  We can optionally employ a collision avoidance technique at this point [13]. This system manages constraints on the human body to prevent body segments from intersecting each other. This process slows the simulation down somewhat, but it is a useful tool when maximum realism is required.

• Step 4: The figure's upper body is now completely postured so we can compute the center of mass. This site is projected onto the ground plane, and compared against the figure's support polygon (Fig. 1). If the projection of this site is outside the support polygon, it is no longer balanced. The active stepping behaviors are used to compute new foot locations that will balance the figure. Leg motions are then executed to place the feet in these new locations [9][5].