Matching Items (5)
Filtering by
- All Subjects: Tolerance (Engineering)
- All Subjects: Computer vision
- Creators: Ren, Yi
Description
Deep neural network-based methods have been proved to achieve outstanding performance on object detection and classification tasks. Deep neural networks follow the ``deeper model with deeper confidence'' belief to gain a higher recognition accuracy. However, reducing these networks' computational costs remains a challenge, which impedes their deployment on embedded devices. For instance, the intersection management of Connected Autonomous Vehicles (CAVs) requires running computationally intensive object recognition algorithms on low-power traffic cameras. This dissertation aims to study the effect of a dynamic hardware and software approach to address this issue. Characteristics of real-world applications can facilitate this dynamic adjustment and reduce the computation. Specifically, this dissertation starts with a dynamic hardware approach that adjusts itself based on the toughness of input and extracts deeper features if needed. Next, an adaptive learning mechanism has been studied that use extracted feature from previous inputs to improve system performance. Finally, a system (ARGOS) was proposed and evaluated that can be run on embedded systems while maintaining the desired accuracy. This system adopts shallow features at inference time, but it can switch to deep features if the system desires a higher accuracy. To improve the performance, ARGOS distills the temporal knowledge from deep features to the shallow system. Moreover, ARGOS reduces the computation furthermore by focusing on regions of interest. The response time and mean average precision are adopted for the performance evaluation to evaluate the proposed ARGOS system.
ContributorsFarhadi, Mohammad (Author) / Yang, Yezhou (Thesis advisor) / Vrudhula, Sarma (Committee member) / Wu, Carole-Jean (Committee member) / Ren, Yi (Committee member) / Arizona State University (Publisher)
Created2022
Description
A process plan is an instruction set for the manufacture of parts generated from detailed design drawings or CAD models. While these plans are highly detailed about machines, tools, fixtures and operation parameters; tolerances typically show up in less formal manner in such plans, if at all. It is not uncommon to see only dimensional plus/minus values on rough sketches accompanying the instructions. On the other hand, design drawings use standard GD&T (Geometrical Dimensioning and tolerancing) symbols with datums and DRFs (Datum Reference Frames) clearly specified. This is not to say that process planners do not consider tolerances; they are implied by way of choices of fixtures, tools, machines, and operations. When converting design tolerances to the manufacturing datum flow, process planners do tolerance charting, that is based on operation sequence but the resulting plans cannot be audited for conformance to design specification.
In this thesis, I will present a framework for explicating the GD&T schema implied by machining process plans. The first step is to derive the DRFs from the fixturing method in each set-up. Then basic dimensions for the features to be machined in each set up are determined with respect to the extracted DRF. Using shop data for the machines and operations involved, the range of possible geometric variations are estimated for each type of tolerances (form, size, orientation, and position). The sequence of manufacturing operations determines the datum flow chain. Once we have a formal manufacturing GD&T schema, we can analyze and compare it to tolerance specifications from design using the T-map math model. Since the model is based on the manufacturing process plan, it is called resulting T-map or m-map. Then the process plan can be validated by adjusting parameters so that the m-map lies within the T-map created for the design drawing. How the m-map is created to be compared with the T-map is the focus of this research.
In this thesis, I will present a framework for explicating the GD&T schema implied by machining process plans. The first step is to derive the DRFs from the fixturing method in each set-up. Then basic dimensions for the features to be machined in each set up are determined with respect to the extracted DRF. Using shop data for the machines and operations involved, the range of possible geometric variations are estimated for each type of tolerances (form, size, orientation, and position). The sequence of manufacturing operations determines the datum flow chain. Once we have a formal manufacturing GD&T schema, we can analyze and compare it to tolerance specifications from design using the T-map math model. Since the model is based on the manufacturing process plan, it is called resulting T-map or m-map. Then the process plan can be validated by adjusting parameters so that the m-map lies within the T-map created for the design drawing. How the m-map is created to be compared with the T-map is the focus of this research.
ContributorsHaghighi, Payam (Author) / Shah, Jami J. (Thesis advisor) / Davidson, Joseph K. (Committee member) / Ren, Yi (Committee member) / Arizona State University (Publisher)
Created2015
Description
When manufacturing large or complex parts, often a rough operation such as casting is used to create the majority of the part geometry. Due to the highly variable nature of the casting process, for mechanical components that require precision surfaces for functionality or assembly with others, some of the important features are machined to specification. Depending on the relative locations of as-cast to-be-machined features and the amount of material at each, the part may be positioned or ‘set up’ on a fixture in a configuration that will ensure that the pre-specified machining operations will successfully clean up the rough surfaces and produce a part that conforms to any assigned tolerances. For a particular part whose features incur excessive deviation in the casting process, it may be that no setup would yield an acceptable final part. The proposed Setup-Map (S-Map) describes the positions and orientations of a part that will allow for it to be successfully machined, and will be able to determine if a particular part cannot be made to specification.
The Setup Map is a point space in six dimensions where each of the six orthogonal coordinates corresponds to one of the rigid-body displacements in three dimensional space: three rotations and three translations. Any point within the boundaries of the Setup-Map (S-Map) corresponds to a small displacement of the part that satisfies the condition that each feature will lie within its associated tolerance zone after machining. The process for creating the S-Map involves the representation of constraints imposed by the tolerances in simple coordinate systems for each to-be-machined feature. Constraints are then transformed to a single coordinate system where the intersection reveals the common allowable ‘setup’ points. Should an intersection of the six-dimensional constraints exist, an optimization scheme is used to choose a single setup that gives the best chance for machining to be completed successfully. Should no intersection exist, the particular part cannot be machined to specification or must be re-worked with weld metal added to specific locations.
The Setup Map is a point space in six dimensions where each of the six orthogonal coordinates corresponds to one of the rigid-body displacements in three dimensional space: three rotations and three translations. Any point within the boundaries of the Setup-Map (S-Map) corresponds to a small displacement of the part that satisfies the condition that each feature will lie within its associated tolerance zone after machining. The process for creating the S-Map involves the representation of constraints imposed by the tolerances in simple coordinate systems for each to-be-machined feature. Constraints are then transformed to a single coordinate system where the intersection reveals the common allowable ‘setup’ points. Should an intersection of the six-dimensional constraints exist, an optimization scheme is used to choose a single setup that gives the best chance for machining to be completed successfully. Should no intersection exist, the particular part cannot be machined to specification or must be re-worked with weld metal added to specific locations.
ContributorsKalish, Nathan (Author) / Davidson, Joseph K. (Thesis advisor) / Shah, Jami J. (Thesis advisor) / Ren, Yi (Committee member) / Arizona State University (Publisher)
Created2016
Description
Tolerance specification for manufacturing components from 3D models is a tedious task and often requires expertise of “detailers”. The work presented here is a part of a larger ongoing project aimed at automating tolerance specification to aid less experienced designers by producing consistent geometric dimensioning and tolerancing (GD&T). Tolerance specification can be separated into two major tasks; tolerance schema generation and tolerance value specification. This thesis will focus on the latter part of automated tolerance specification, namely tolerance value allocation and analysis. The tolerance schema (sans values) required prior to these tasks have already been generated by the auto-tolerancing software. This information is communicated through a constraint tolerance feature graph file developed previously at Design Automation Lab (DAL) and is consistent with ASME Y14.5 standard.
The objective of this research is to allocate tolerance values to ensure that the assemblability conditions are satisfied. Assemblability refers to “the ability to assemble/fit a set of parts in specified configuration given a nominal geometry and its corresponding tolerances”. Assemblability is determined by the clearances between the mating features. These clearances are affected by accumulation of tolerances in tolerance loops and hence, the tolerance loops are extracted first. Once tolerance loops have been identified initial tolerance values are allocated to the contributors in these loops. It is highly unlikely that the initial allocation would satisfice assemblability requirements. Overlapping loops have to be simultaneously satisfied progressively. Hence, tolerances will need to be re-allocated iteratively. This is done with the help of tolerance analysis module.
The tolerance allocation and analysis module receives the constraint graph which contains all basic dimensions and mating constraints from the generated schema. The tolerance loops are detected by traversing the constraint graph. The initial allocation distributes the tolerance budget computed from clearance available in the loop, among its contributors in proportion to the associated nominal dimensions. The analysis module subjects the loops to 3D parametric variation analysis and estimates the variation parameters for the clearances. The re-allocation module uses hill climbing heuristics derived from the distribution parameters to select a loop. Re-allocation Of the tolerance values is done using sensitivities and the weights associated with the contributors in the stack.
Several test cases have been run with this software and the desired user input acceptance rates are achieved. Three test cases are presented and output of each module is discussed.
The objective of this research is to allocate tolerance values to ensure that the assemblability conditions are satisfied. Assemblability refers to “the ability to assemble/fit a set of parts in specified configuration given a nominal geometry and its corresponding tolerances”. Assemblability is determined by the clearances between the mating features. These clearances are affected by accumulation of tolerances in tolerance loops and hence, the tolerance loops are extracted first. Once tolerance loops have been identified initial tolerance values are allocated to the contributors in these loops. It is highly unlikely that the initial allocation would satisfice assemblability requirements. Overlapping loops have to be simultaneously satisfied progressively. Hence, tolerances will need to be re-allocated iteratively. This is done with the help of tolerance analysis module.
The tolerance allocation and analysis module receives the constraint graph which contains all basic dimensions and mating constraints from the generated schema. The tolerance loops are detected by traversing the constraint graph. The initial allocation distributes the tolerance budget computed from clearance available in the loop, among its contributors in proportion to the associated nominal dimensions. The analysis module subjects the loops to 3D parametric variation analysis and estimates the variation parameters for the clearances. The re-allocation module uses hill climbing heuristics derived from the distribution parameters to select a loop. Re-allocation Of the tolerance values is done using sensitivities and the weights associated with the contributors in the stack.
Several test cases have been run with this software and the desired user input acceptance rates are achieved. Three test cases are presented and output of each module is discussed.
ContributorsBiswas, Deepanjan (Author) / Shah, Jami J. (Thesis advisor) / Davidson, Joseph (Committee member) / Ren, Yi (Committee member) / Arizona State University (Publisher)
Created2016
Description
Autonomous Driving (AD) systems are being researched and developed actively in recent days to solve the task of controlling the vehicles safely without human intervention. One method to solve such task is through deep Reinforcement Learning (RL) approach. In deep RL, the main objective is to find an optimal control behavior, often called policy performed by an agent, which is AD system in this case. This policy is usually learned through Deep Neural Networks (DNNs) based on the observations that the agent perceives along with rewards feedback received from environment.However, recent studies demonstrated the vulnerability of such control policies learned through deep RL against adversarial attacks. This raises concerns about the application of such policies to risk-sensitive tasks like AD. Previous adversarial attacks assume that the threats can be broadly realized in two ways: First one is targeted attacks through manipu- lation of the agent’s complete observation in real time and the other is untargeted attacks through manipulation of objects in environment. The former assumes full access to the agent’s observations at almost all time, while the latter has no control over outcomes of attack. This research investigates the feasibility of targeted attacks through physical adver- sarial objects in the environment, a threat that combines the effectiveness and practicality.
Through simulations on one of the popular AD systems, it is demonstrated that a fixed optimal policy can be malfunctioned over time by an attacker e.g., performing an unintended self-parking, when an adversarial object is present. The proposed approach is formulated in such a way that the attacker can learn a dynamics of the environment and also utilizes common knowledge of agent’s dynamics to realize the attack. Further, several experiments are conducted to show the effectiveness of the proposed attack on different driving scenarios empirically. Lastly, this work also studies robustness of object location, and trade-off between the attack strength and attack length based on proposed evaluation metrics.
ContributorsBuddareddygari, Prasanth (Author) / Yang, Yezhou (Thesis advisor) / Ren, Yi (Committee member) / Fainekos, Georgios (Committee member) / Arizona State University (Publisher)
Created2021