Using semantic Geometric Dimensioning and Tolerancing (GD &T) information from STEP AP242 neutral exchange files for robotic applications

The two essential parameters of any insertion operation are clearance and insertion length. Clearance is also represented as the clearance ratio, and it is defined as the ratio of the clearance to hole diameter [20]:where \(r_c\) is the clearance ratio, \(d_h\) is the hole diameter, and \(d_p\) is the peg diameter.

Part clearances, clearance ratio, and insertion lengths can be used to estimate the contact states and force needed to complete the assembly. The clearance ratio also determines the maximum permissible tilt during the assembly process. Simunovic [21] used the part dimensions and insertion lengths to estimate the type of contact and insertion forces during assembly. Whitney [20] used the clearance ratio and insertion lengths to estimate the contact states, forces and jamming during peg-in-hole assembly. Clearance also determines the maximum tilt and the conditions of jamming in a chamferless assembly as described by Haskiya et al. [22]. The applicability of contact and force models also depends on the part clearances [23].

Part clearances also determine the suitability of the robotic manipulator and the success of the assembly operation. It also determines the control methodology needed for completion of the assembly [24, 25]. ElMarghy et al. [26] discussed the effects of part dimensions and tolerances along with manipulator accuracy and repeatability on the success of assembly operation. As the uncertainty-to-clearance ratio increases, the completion of assembly becomes difficult with position control alone [27, 28]. Hence depending on the clearance, an appropriate control strategy can be selected [29]. As the clearance ratio decreases, the success of assembly decreased [28, 30].

The extraction of part dimensions and tolerances from the STEP files is used to calculate the effective clearance, which can be used to estimate the contact states and possibility of jamming; the assembly forces needed; decide the suitability of a manipulator for the assembly task; and to decide the control strategy.

Once the toleranced features and the tolerance values are extracted, these values are used to calculate the variation of these features from the nominal values. These values are used for defining the constraints for the robotic assembly tasks. Automated tolerance analysis programs can calculate the variation of toleranced features and their propagation in the entire assembly. Effective (worst possible) clearances between mating parts can be evaluated using these tolerance values. Considering the size tolerances, the effective clearance can be calculated as [26]where \(c_e\) is the effective clearance, \(t_h\) is the tolerance on hole diameter, and \(t_p\) is the tolerance on peg diameter. Using Monte-Carlo simulations and other automated computer programs, the tolerance information can be analyzed to describe the clearances that can be expected.

Many designs have multiple features that can come into contact during the assembly operation. Providing the complete design with PMI makes it possible to evaluate the assembly process to determine which features are relevant to the process. For robot programs, the change in relevant features can require a change in control strategy required depending on the tolerance values.

PMI in general will be very useful in the automation of welding process. The PMI along with product geometry can be used for automatic programming of welding robots. This will decrease the dependence on costly and time consuming manual programming and increases its adoption in SMEs where changing product design requires frequent changes in robot programs. Mohammed et al. [31] discussed the importance of PMI in robotic welding and described the reuse of welding information for STEP AP242 files. In addition to welding annotations, GD &T information will also be useful in automated robotic welding.

The quality and the success of the weld depends on the root gap and the fit–up of the components being welded. These two parameters depend on the seam geometry, edge preparation and material thickness. The geometric variations of the parts being welded also effect the seam position, orientation and geometry, thus affecting the weld path, positioning and orientations of weld gun. The effects of these variations are limited by specifying various tolerances. The tolerances that impact the weld quality can be divided into two groups.In manual welding, the welder senses the variations and errors and makes adjustments to successfully weld the parts. The product variations should be taken into account in robotic welding processes to avoid product quality issues and rejections. Generally in robotic welding, sensors are used to scan and extract the seam and groove geometry [32]. The sensing operation is planned based on the human input. This can be improved using the information from the CAD files. This paper demonstrates how the groove tolerances can be included in the welding annotations.

A STEP file is a text-based file that encapsulates all the product details in entities. These files are called Part-21 files because this text encoding is done as per Part-21 of ISO 10303 standard [46]. These files are also known as physical files and have the extension ‘.STP’ or ‘.STEP’. These files are human-readable and facilitate the writing, reading, and exchange of product data. The significant entities used for representing product geometric information are

Figure 2 shows the different types of entities and their relationships for a rectangular block. The geometric entities represent the basic geometric elements of product design like a Cartesian point, a line, or a plane. These entities are combined or referred to in topological entities to form the features of the product geometry like a vertex, an edge or a face of the product. The higher entities bind all the topological entities and represent the entire product, for example, the ‘Mechanical_Design_Geometric_Presentation_Representation’ entity.

Where possible, the annotations are attached to the topological entities to refer to the part features that are affected by these annotations. This is also suggested by the CAx–IF recommendation [47], where it recommends using the ‘advanced_face’ entity for part features while attaching the GD &T annotations. This helps identify the part features quickly and enhances the reuse of product data in the downstream operations.

The GD &T information and feature control frames can be included in the STEP files using Unicode strings based on the CAx recommendations of PMI Unicode String Specification Examples and Mapping Strategies [53]. All the GD &T symbols defined in ISO and ASME standards are represented using Unicode characters. Table 2 shows the Unicode and custom codes of GD &T symbols.

All the GD &T information can be included in the STEP AP242 files using the entities defined in the standard. The methodology followed is given in CAx recommended practices for ”Representation and Presentation of Product Manufacturing Information (PMI) (AP242)” [47]. From this recommendation, only ’representation’ aspects are relevant as the information is semantically represented and reusable for robotic applications. This CAx recommendation uses the entities defined in the first edition of STEP AP242.

The authors of this article have analyzed the second edition of STEP AP242 and performed a comparative study against the first edition for the present work. The second edition has a set of changes in type and entity definitions from the first edition. However, these changes do not affect the methodology described in the CAx recommendation. Hence, the authors suggest using the same methodology described by CAx using the second edition of STEP AP242. The entities from the following Application Modules (AM) can capture the GD &T information of the parts: Dimension tolerance (Part: 1050), Geometric tolerance (Part: 1051), and Default tolerance (Part: 1052).

The entities from the first three AMs are mainly mapped to the entities of Integrated Generic Resource: Shape variation tolerances (Part: 47). The Part: 47 has three schemas that cover the information requirements for GD &T defined in these AMs. The schemas of Part: 47 are ‘Shape aspect definition, Shape dimension and Shape tolerance’. Using the entities from these AMs and following the methodology of CAx recommendation, all the GD &T information can be added semantically to the part features.

Another useful AM is ‘Extended geometric tolerance’ (Part:1666). This AM specifies the entities for defining the tolerance zone boundaries and can be used for automated tolerance analysis and quality analysis.

Extraction of GD &T information from the Unicode string is straightforward. From the ‘SHAPE_REPRESENTATION’ entity, the part feature and the corresponding Unicode tolerance annotation can be identified. The feature geometry is extracted using the standard geometric and topological entities of STEP AP242 standard, and the Unicode string is extracted from the entity ‘TEXT_LITERAL’. A parser is developed to parse the Unicode annotation and extract the tolerance information. This information is used in forming the tolerance zones and tolerance analysis.

The overall process for extracting the tolerance information while using the standard entities from STEP AP242 requires more steps than the Unicode string. A flowchart of the process is shown in Fig. 5. After reading the STEP file, the program searches and identifies the tolerance entities. These are the entities from the schemas of Shape variation tolerances (Part: 47) of the standard. These entities are checked to determine whether these tolerances are connected to the geometry of part features. If the tolerance entities refer to ‘SHAPE_ASPECT’ or ‘COMPOSITE_GROUP_SHAPE_ASPECT,’ then these tolerances are attached to the part features. This corresponds to the path ‘yes’ after the ‘Connected to Geometry’ check in Fig. 5. In this case, the corresponding part feature and its geometry are extracted from the topological and geometric entities.

The default tolerances are not attached to any particular geometric feature, hence do not refer to ‘SHAPE_ASPECT’ or ‘COMPOSITE_GROUP_SHAPE_ASPECT’ entities.This corresponds to the path ‘no’ after the ‘Connected to Geometry’ check in Fig. 5. These default tolerances apply to all the dimensions for which specific tolerances are not given. Traditionally these tolerances are shown in the ‘title block’ of the manufacturing drawing. The type and values of default tolerances are extracted from the entities of Part: 1052 (Default tolerances). The values of these default tolerances are then passed on to the tolerance analysis programs.

After extracting the topological and geometric details of the entities, the attached dimension and tolerance information is extracted. The tolerance information is inferred from various entities from the schemas of Part: 47 (Shape variation tolerances). Figure 6 shows the important steps in this process. First, it is determined whether the annotation is a dimension or has a datum or an FCF. If a dimension is attached to the feature, then the type of the dimension, i.e., dimension of size or location or distance or angle, is determined. Then the nominal value is identified. The type and value of tolerance affecting this dimension are determined. The dimensions and tolerance values from the standard entities can be readily extracted from the entities corresponding to Part: 1050 (Dimensions tolerances).

If a datum callout is attached, the name is identified and attached to the part feature. The datum target values are not considered in this paper. In the case of an FCF, all the values related to the geometric tolerances along with the modifiers are extracted as shown in Fig. 6. These values can be extracted from the entity mappings defined in Part: 1051 (Geometric tolerances). Once all the information is extracted, it is linked to the corresponding part feature. The tolerance information and the corresponding feature geometry are further used to define the tolerance zones and perform tolerance analysis, described in the next section.

This process is demonstrated using a motor assembly design provided by Mjøs Metallvare shown in Fig. 7. The tolerances applied to these three major motor sub-assemblies are also shown. The tolerance values shown are created for demonstration purposes and do not correspond with the true tolerance values in the product. Figure 8 shows the portion of a STEP file with standards entities adding a flatness tolerance and a datum callout attached to the FCF of the flatness tolerance. The clearance ratios between the Rotor-Housing and between Rotor-EndPlate are calculated using these tolerance values. The overall process for calculating the effective clearance from the mating features and their tolerances is shown in Fig. 9. The lowest clearance ratio value is considered for programming of the assembly processes where mating/insertion co-occurs at two features. In the present use-case, the insertion/mating happens at two features simultaneously during the assembly of Rotor with Housing. The different stages of this assembly are shown in Fig. 10. The start and end of these two stages of mating can be estimated from the insertion lengths, and the insertion lengths are calculated from the mating constraints and the tolerance values. In this case, the overall insertion length is from the bottom of the bearing seat to its corresponding mating surface on the bearing, approximately 162 mm. The Rotor reaches the Housing in the initial approach and achieves rough alignment. During this stage, the manipulator has much freedom in angle and position. Precise control is unnecessary, and the manipulator path can be planned based on position control.

In the first stage, the first possible contact situation occurs between the magnet region of the Rotor with the windings in the Housing. At the beginning of this stage, the clearance between the components is about 3 mm, and it reduces to 0.25 mm as the insertion proceeds. Precise position control is needed from this stage.

The second stage starts when the bearing starts mating with the bearing seat in the bottom plate of the Housing sub-assembly. This stage starts after insertion of about 142 mm and ends after another 19 mm insertion. The clearance between the housing seat and the bearing outer diameter is around 0.03 mm, requiring sensor-feedback control.

The assembly task is completed when the bearing is seated in the Housing. Successful assembly is achieved when a sufficient force is detected and the Rotor is placed within the expected region identified by tolerance analysis.

Similarly, the assembly of EndPlate with the rest of the motor can be divided into first stage when it starts mating with the shaft and the second stage when the top bearing starts mating with the bearing housing on the EndPlate. In this step, the first stage has very high clearance, and the clearances in the second stage are the same as that of the second stage of the earlier step.

The same process is applied to the T–joint of two pipes shown in Fig. 11. In case of structural welds involving beams, the weld seams are along the edges of mating surfaces. These can be directly identified and extracted from the CAD models. Extraction and identification of welds involving cylindrical parts like pipe welds are a bit complicated. Generally welders cut the profiles in the pipes before welding using wrap-around template curves. When these weld joints are properly modelled in the CAD files, then the seam curves will be readily available in the STEP files. The ‘B_SPLINE_CURVE_WITH_KNOTS’ entity from the STEP AP242 file shown in Fig. 12 gives the weld seam of the T-joint shown in Fig. 11. From this, the nominal geometry of the seam curve was derived. The information from GD &T annotations is used to calculate the deviations from the nominal geometry. The identified weld seams will be in the local coordinate frame of the mating features. These have to be represented with respect to the coordinate frame of the welding cell in which the weld torch is described.

The Unicode string for welding annotation is included as ‘TEXT_LITERAL’ entity. The groove geometry was extracted from the weld annotations. The weld groove is bevel with a bevel angle of \(37.5^{\circ } {\pm } 2.5^{\circ }\) and root face of 1.6 ± 0.4 mm. The Unicode weld string is modified by adding the tolerance value after the nominal value.The groove geometry with the seam can be used as an input for defining the sensing window for planning the scanning using a visual sensor system.