PCM: Automatic Clustering Method for Real-time Construction Simulation
即時模擬施工使用的自動集群方法: PCM
PCM: Automatic Clustering Method for Real-time Construction Simulation
即時模擬施工使用的自動集群方法: PCM
Related references
1. W. H. Hung and S. C. Kang, "Automatic Clustering Method for Real-time Construction Simulation," Advanced Engineering Informatics, reviewing. 2. H. L. Chi, W. H. Hung, S. C. Kang, A Physics Based Simulation for Crane Manipulation and Cooperation, Computing in Civil Engineering Conference, 2007. 3. J. R. Juang, W. H. Hung, S. C. Kang, Using Game Engines for Physical– Based Simulations – A Forklift, Electronic J. Inform. Tech. Constr. 16 (2011) 3–22. 4. W. H. Hung, S. C. Kang, Physics-based Crane Model for the Simulation of Cooperative Erections, Proceedings of the 9th international confer- ence on Construction Applications of Virtual Reality, 2009. 5. C. E. Yang, J. C. Ling, W. H. Hung, S. C. Kang, Accessibility Evaluation System for Site Layout Planning, in: Proceedings of 10th International Conference on Construction Applications of Virtual Reality (CONVR) Conference, 2010. 6. K. C. Lai, S. C. Kang, Collision Detection Strategies for Virtual Con- struction Simulation, Autom. Constr. 18 (6) (2009) 724–736. 7. H. L. Chi, S. C. Kang, A Physics-based Simulation Approach for Coop- erative Erection Activities, Autom. Constr. 19 (6) (2010
PCM: Automatic Clustering Method for Real-time Construction Simulation 即時模擬施工使用的自動集群方法: PCM 在虛擬實境中模擬建築工程可以防止施工過程中可能發生的意外,提高施工現場的效率及安全。一般在 即時模擬的過程中,需要在1/10或1/20秒的時間內計算出物體間是否有碰撞,這些物體可能有數千到數 百萬個,這在計算上造成了很大的困難。因此,降低運算量成為最重要的一件事。一個常見且有效的方 法是將好幾個物體集成一群,用一個較大的碰撞檢測邊界表示這幾個物體,這可以有效的降低所需要的 運算量。然而,在一個較大的施工情境下,用人工去將這些物體集群是非常花時間的。 此研究提出了一個方法, Propagation Clustering Method(PCM),此方法使用k-means演算法,可以 在即時的狀況下自動產生一個碰撞檢測邊界。使用者可以藉由改變權重來調整適合的集群結果,此方法 也定義了一個指標來評估產生的碰撞檢測邊界。該方法還可以產生多層的碰撞檢測邊界,可以在不同層 次的細節進行碰撞檢測,保留在碰撞檢測的效率和準確性。 為了驗證此方法可行性,我們定義了三個測試場景:一個有許多物體的工地,如一個小廠的建設;高樓 層建築的工地,及一個有許多物體,又是高樓層建築的大型工地。實驗結果表明,PCM可以有效的將物 體集群。此方法顯著的降低模擬施工所需的運算量。
The workflow of the proposed method Models
Append i - depth Nodes
AABB Generation
Model Clustering
Object Groups
i =0 Object Dataset
Hierarchical Boundary Tree
Re-cluster the unsatisfied groups no
i = i +1 Quality Evaluation (are all groups satisfy the criteria)
yes
Boundary Generation
PCM: Automatic Clustering Method for Real-time Construction Simulation
Simulation of construction activities in a virtual environment can prevent constructability problems and increase efficiency and safety at the physical construction site. The computation for collision checks creates a bottleneck during these simulations. A typical construction simulation requires collision checks to be performed between all pairs among thousands or even millions of objects, and each of these checks must be completed within 1/10th or even 1/20th of a second to provide a smooth real-time simulation. Therefore, the reduction of computational cost is paramount. An effective and commonly used method is to cluster the objects into groups and use a larger surrounding boundary shape in place of the individual objects. This significantly reduces the computational effort required. However, clustering objects manually is usually time consuming and may be impossible for large scenarios. This paper presented a method, called the Propagation Clustering Method (PCM), for automatically generating the collision detection boundaries of a construction site model in real-time construction simulations. The proposed PCM uses a modified version of the k-means clustering method to cluster the objects in a construction site model into groups iteratively. Each object in a construction site model is represented by an axis-aligned bounding box (AABB) bounding shape as a six-dimensional data point. The method can thus classify the objects by their dimensions and positions. The weighted coefficients then allow users to adjust and tune the clustering result for different kinds of simulations. This research also defined an evaluation index for the clustered results by considering the volume of the generated collision detection boundaries and the objects’ AABBs. Developers can utilize the proposed method to generate appropriate collision detection boundaries quickly, and with different efficiencies and accuracies for different simulation purposes. The proposed method can also generate a hierarchical collision detection boundary tree. By performing collision detection in the tree to different depths and levels of the detail, the hierarchical boundary tree can retain both efficiency and accuracy in collision detection.
The workflow of the PCM module.� Inputs
outputs
PCM Module
Model
Hierarchical Boundary Tree
Dynamic Boundary Generator
Dynamic Boundary
Model Groups
Static Boundary Generator
Static Boundary
Generate bounding box dataset
Model Dataset
Clustering Parameters
PCM Cluster
PCM: Automatic Clustering Method for Real-time Construction Simulation Using k-means clustering method: (a) original model; (b) clustering without weights; (c) clustering with weights where (wx , wy , wz , ww , wh , wl ) = (3,1,3,1,1,1)
(a)
(b)
(c)
Collision detection boundary constructed using PCM with dierent number of iterations. Iteration = 1
Iteration = 2
Iteration = 3
Iteration = 4
Iteration = 5
Iteration = 6
Real-time visualization of the collision detection in scenario 2 by using the proposed hierarchical collision detection boundary tree.
y x
Time: 0
z
Time: 5 collided object
Time: 10
Time: 15
Shih-Chung Jessy Kang sckang@ntu.edu.tw sckang.caece.net