Design Article
Enhancing simulation studies with 3D animation
Saurabh Mahapatra, MathWorks
6/8/2011 5:51 AM EDT
Abstract
The use of simulation studies to better understand the dynamic behavior of a system under investigation is at the core of verifying designs early in the development process. Despite the amount of data that such studies produce, a 3D representation of the system creates a more complete understanding of system behavior. This article describes the use of 3D animation in simulation-centric workflows to augment early verification activities, such as those used in Model-Based Design. The evolution of technology and domain specialization in the simulation and 3D graphics fields presents several challenges for using 3D animation in simulation-centric studies. A set of examples illustrates how to meet these challenges.
Harvesting data from simulations
The last decade has seen the increasing use of computer technology to prototype engineering designs before hardware manufacture or deployment. One area that has generated significant interest is the study of dynamic systems in which nonlinear equations determine the governing dynamics. Providing the means of representing these systems as software programs enables multiple simulations to be run based on various inputs to the system. This methodology presents opportunities for the simulation engineer, such as the ability to vary system parameters or environmental inputs to reflect possible-use scenarios, and to do so exhaustively. Early verification through simulations minimizes the risk inherent in the design process by reducing the probability of discovering errors late in the development process.
Effective early verification requires the careful analysis of simulation data to enable the understanding of system behavior. For example, an engineer could analyze data sets to discover patterns of unintended behaviors resulted during multiple runs. The analysis of such data, however, can pose a challenge as it creates the need for domain-specific experts who decipher it.
Towards a better understanding of multidimensional simulation data
What makes the understanding of simulation data from a dynamic system difficult? Typically, it is the interdependencies in the multidimensional data caused by the equations describing the system. In addition to the mathematical analyses that can be done on raw data, visual representations of the data such as 2D and 3D plots for cluster or trend analysis would enrich our understanding of the data relationships due to their placement relative to each other or with respect to a parameter, such as time.
Despite the obvious advantages of using the analytic and visual methods together, this approach would provide some challenges to our understanding in specific situations, especially when the dimensionality of the data is large. Consider the simple physics of an aircraft whose physics modeled a rigid body having six degrees of freedom (6DoF). To analyze the motion of this aircraft, the dimension of the data in time would be six, representing three positional and three rotational coordinates. Besides the obvious limitations in deciphering trends in the raw data, graphical representations such as 2D and 3D plots will be insufficient to improve our understanding of the rigid body motion in this six-dimensional data space. The task becomes harder if we add dimensional complexity to this example by measuring temperature data in time as a function of surface geometry.
Much has been written in 3D graphics literature about a picture being worth a thousand words. Some authors have ambitiously extrapolated that moving images are worth even more. Over the decades, the success of the 3D video game industry and advances in 3D graphics hardware and software lend credibility to this observation. But from a simulation-centric standpoint, data representation using 3D scene representations is promising, if the system under study has interesting interdependencies that can be perceived by the analyst. For example, a 3D animation of multiple agents collaborating to achieve a common or different goal would provide insights in the area of swarm studies. In general, dynamic simulations with their associated physics, such as mechanics, thermodynamics, acoustics or multibody interaction present opportunities for visualization through 3D animation.
A combined strategy that leverages the above approaches can enhance early verification. Figure 1 shows an example of vehicle dynamics simulation in Simulink® as a hybrid representation that combines the three approaches. The dimensionality of the data associated with this simulation is 78—mechanical parameters of the 4 wheels (64) each consisting of positions (3), rotations of wheel axis (9), spin (1), and forces (3). The remaining (14) are associated with the braking signal (1), steering input signal (1), car position (3), and car rotation (9).
Figure 1: Hybrid representation that combines 1D raw data, 2D graphs, and 3D animation approaches.
The use of simulation studies to better understand the dynamic behavior of a system under investigation is at the core of verifying designs early in the development process. Despite the amount of data that such studies produce, a 3D representation of the system creates a more complete understanding of system behavior. This article describes the use of 3D animation in simulation-centric workflows to augment early verification activities, such as those used in Model-Based Design. The evolution of technology and domain specialization in the simulation and 3D graphics fields presents several challenges for using 3D animation in simulation-centric studies. A set of examples illustrates how to meet these challenges.
Harvesting data from simulations
The last decade has seen the increasing use of computer technology to prototype engineering designs before hardware manufacture or deployment. One area that has generated significant interest is the study of dynamic systems in which nonlinear equations determine the governing dynamics. Providing the means of representing these systems as software programs enables multiple simulations to be run based on various inputs to the system. This methodology presents opportunities for the simulation engineer, such as the ability to vary system parameters or environmental inputs to reflect possible-use scenarios, and to do so exhaustively. Early verification through simulations minimizes the risk inherent in the design process by reducing the probability of discovering errors late in the development process.
Effective early verification requires the careful analysis of simulation data to enable the understanding of system behavior. For example, an engineer could analyze data sets to discover patterns of unintended behaviors resulted during multiple runs. The analysis of such data, however, can pose a challenge as it creates the need for domain-specific experts who decipher it.
Towards a better understanding of multidimensional simulation data
What makes the understanding of simulation data from a dynamic system difficult? Typically, it is the interdependencies in the multidimensional data caused by the equations describing the system. In addition to the mathematical analyses that can be done on raw data, visual representations of the data such as 2D and 3D plots for cluster or trend analysis would enrich our understanding of the data relationships due to their placement relative to each other or with respect to a parameter, such as time.
Despite the obvious advantages of using the analytic and visual methods together, this approach would provide some challenges to our understanding in specific situations, especially when the dimensionality of the data is large. Consider the simple physics of an aircraft whose physics modeled a rigid body having six degrees of freedom (6DoF). To analyze the motion of this aircraft, the dimension of the data in time would be six, representing three positional and three rotational coordinates. Besides the obvious limitations in deciphering trends in the raw data, graphical representations such as 2D and 3D plots will be insufficient to improve our understanding of the rigid body motion in this six-dimensional data space. The task becomes harder if we add dimensional complexity to this example by measuring temperature data in time as a function of surface geometry.
Much has been written in 3D graphics literature about a picture being worth a thousand words. Some authors have ambitiously extrapolated that moving images are worth even more. Over the decades, the success of the 3D video game industry and advances in 3D graphics hardware and software lend credibility to this observation. But from a simulation-centric standpoint, data representation using 3D scene representations is promising, if the system under study has interesting interdependencies that can be perceived by the analyst. For example, a 3D animation of multiple agents collaborating to achieve a common or different goal would provide insights in the area of swarm studies. In general, dynamic simulations with their associated physics, such as mechanics, thermodynamics, acoustics or multibody interaction present opportunities for visualization through 3D animation.
A combined strategy that leverages the above approaches can enhance early verification. Figure 1 shows an example of vehicle dynamics simulation in Simulink® as a hybrid representation that combines the three approaches. The dimensionality of the data associated with this simulation is 78—mechanical parameters of the 4 wheels (64) each consisting of positions (3), rotations of wheel axis (9), spin (1), and forces (3). The remaining (14) are associated with the braking signal (1), steering input signal (1), car position (3), and car rotation (9).
Figure 1: Hybrid representation that combines 1D raw data, 2D graphs, and 3D animation approaches.
|
|
|
|
|
Navigate to related information
