Matching Items (131)
Description
Physical and structural tree measurements are applied in forestry, precision agriculture and conservation for various reasons. Since measuring tree properties manually is tedious, measurements from only a small subset of trees present in a forest, agricultural land or survey site are often used. Utilizing robotics to autonomously estimate physical tree dimensions would speed up the measurement or data collection process and allow for a much larger set of trees to be used in studies. In turn, this would allow studies to make more generalizable inferences about areas with trees. To this end, this thesis focuses on developing a system that generates a semantic representation of the topology of a tree in real-time. The first part describes a simulation environment and a real-world sensor suite to develop and test the tree mapping pipeline proposed in this thesis. The second part presents details of the proposed tree mapping pipeline. Stage one of the mapping pipeline utilizes a deep learning network to detect woody and cylindrical portions of a tree like trunks and branches based on popular semantic segmentation networks. Stage two of the pipeline proposes an algorithm to separate the detected portions of a tree into individual trunk and branch segments. The third stage implements an optimization algorithm to represent each segment parametrically as a cylinder. The fourth stage formulates a multi-sensor factor graph to incrementally integrate and optimize the semantic tree map while also fusing two forms of odometry. Finally, results from all the stages of the tree mapping pipeline using simulation and real-world data are presented. With these implementations, this thesis provides an end-to-end system to estimate tree topology through semantic representations for forestry and precision agriculture applications.
ContributorsVishwanatha, Rakshith (Author) / Das, Jnaneshwar (Thesis advisor) / Martin, Roberta (Committee member) / Throop, Heather (Committee member) / Zhang, Wenlong (Committee member) / Ehsani, Reza (Committee member) / Arizona State University (Publisher)
Created2022
Description
Autonomous systems inevitably must interact with other surrounding systems; thus, algorithms for intention/behavior estimation are of great interest. This thesis dissertation focuses on developing passive and active model discrimination algorithms (PMD and AMD) with applications to set-valued intention identification and fault detection for uncertain/bounded-error dynamical systems. PMD uses the obtained input-output data to invalidate the models, while AMD designs an auxiliary input to assist the discrimination process. First, PMD algorithms are proposed for noisy switched nonlinear systems constrained by metric/signal temporal logic specifications, including systems with lossy data modeled by (m,k)-firm constraints. Specifically, optimization-based algorithms are introduced for analyzing the detectability/distinguishability of models and for ruling out models that are inconsistent with observations at run time. On the other hand, two AMD approaches are designed for noisy switched nonlinear models and piecewise affine inclusion models, which involve bilevel optimization with integer variables/constraints in the inner/lower level. The first approach solves the inner problem using mixed-integer parametric optimization, whose solution is included when solving the outer problem/higher level, while the second approach moves the integer variables/constraints to the outer problem in a manner that retains feasibility and recasts the problem as a tractable mixed-integer linear programming (MILP). Furthermore, AMD algorithms are proposed for noisy discrete-time affine time-invariant systems constrained by disjunctive and coupled safety constraints. To overcome the issues associated with generalized semi-infinite constraints due to state-dependent input constraints and disjunctive safety constraints, several constraint reformulations are proposed to recast the AMD problems as tractable MILPs. Finally, partition-based AMD approaches are proposed for noisy discrete-time affine time-invariant models with model-independent parameters and output measurement that are revealed at run time. Specifically, algorithms with fixed and adaptive partitions are proposed, where the latter improves on the performance of the former by allowing the partitions to be optimized. By partitioning the operation region, the problem is solved offline, and partition trees are constructed which can be used as a `look-up table' to determine the optimal input depending on revealed information at run time.
ContributorsNiu, Ruochen (Author) / Yong, Sze Zheng S.Z. (Thesis advisor) / Berman, Spring (Committee member) / Ren, Yi (Committee member) / Zhang, Wenlong (Committee member) / Zhuang, Houlong (Committee member) / Arizona State University (Publisher)
Created2022
Description
Adversarial threats of deep learning are increasingly becoming a concern due to the ubiquitous deployment of deep neural networks(DNNs) in many security-sensitive domains. Among the existing threats, adversarial weight perturbation is an emerging class of threats that attempts to perturb the weight parameters of DNNs to breach security and privacy.In this thesis, the first weight perturbation attack introduced is called Bit-Flip Attack (BFA), which can maliciously flip a small number of bits within a computer’s main memory system storing the DNN weight parameter to achieve malicious objectives. Our developed algorithm can achieve three specific attack objectives: I) Un-targeted accuracy degradation attack, ii) Targeted attack, & iii) Trojan attack. Moreover, BFA utilizes the rowhammer technique to demonstrate the bit-flip attack in an actual computer prototype. While the bit-flip attack is conducted in a white-box setting, the subsequent contribution of this thesis is to develop another novel weight perturbation attack in a black-box setting. Consequently, this thesis discusses a new study of DNN model vulnerabilities in a multi-tenant Field Programmable Gate Array (FPGA) cloud under a strict black-box framework. This newly developed attack framework injects faults in the malicious tenant by duplicating specific DNN weight packages during data transmission between off-chip memory and on-chip buffer of a victim FPGA. The proposed attack is also experimentally validated in a multi-tenant cloud FPGA prototype. In the final part, the focus shifts toward deep learning model privacy, popularly known as model extraction, that can steal partial DNN weight parameters remotely with the aid of a memory side-channel attack. In addition, a novel training algorithm is designed to utilize the partially leaked DNN weight bit information, making the model extraction attack more effective. The algorithm effectively leverages the partial leaked bit information and generates a substitute prototype of the victim model with almost identical performance to the victim.
ContributorsRakin, Adnan Siraj (Author) / Fan, Deliang (Thesis advisor) / Chakrabarti, Chaitali (Committee member) / Seo, Jae-Sun (Committee member) / Cao, Yu (Committee member) / Arizona State University (Publisher)
Created2022
Description
Artificial Intelligence (AI) and Machine Learning (ML) techniques have come a long way since their inception and have been used to build intelligent systems for a wide range of applications in everyday life. However they are very computationintensive and require transfer of large volume of data from memory to the computation units. This memory access time constitute significant part of the computational latency and a performance bottleneck. To address this limitation and the ever-growing demand for implementation in hand-held and edge-devices, In-memory computing (IMC) based AI/ML hardware accelerators have emerged. First, the dissertation presents an IMC static random access memory (SRAM) based hardware modeling and optimization framework. A unified systematic study closely models the IMC hardware, and investigates how a number of design variables and non-idealities (e.g. device mismatch and ADC quantization) affect the Deep Neural Network (DNN) accuracy of the IMC design. The framework allows co-optimized selection of different design variables accounting for sources of noise in IMC hardware and robust implementation of a high accuracy DNN. Next, it presents a kNN hardware accelerator in 65nm Complementary Metal-Oxide-Semiconductor (CMOS) technology. The accelerator combines an IMC SRAM that is developed for binarized deep neural networks and other digital hardware that performs top-k sorting. The simulated k Nearest Neighbor accelerator design processes up to 17.9 million query vectors per second while consuming 11.8 mW, demonstrating >4.8× energy-efficiency improvement over prior works. This dissertation also presents a novel floating-point precision IMC (FP-IMC) macro with a hybrid architecture that configurably supports two Floating Point (FP) precisions. Implementing FP precision MAC has been a challenge owing to its complexity. The design is implemented on 28nm CMOS, and taped-out on chip demonstrating 12.1 TFLOPS/W and 66.1 TFLOPS/W for 8-bit Floating Point (FP8) and Block Floating point (BF8) respectively. Finally, another iteration of the FP design is presented that is modeled to support multiple precision modes from FP8 up to FP32. Two approaches to the architectural design were compared illustrating the throughput-area overhead trade-off. The simulated design shows a 2.1 × normalized energy-efficiency compared to the on-chip implementation of the FP-IMC.
ContributorsSaikia, Jyotishman (Author) / Seo, Jae-Sun (Thesis advisor) / Chakrabarti, Chaitali (Thesis advisor) / Fan, Deliang (Committee member) / Cao, Yu (Committee member) / Arizona State University (Publisher)
Created2023
Description
Soft robotics has garnered attention for its substantial prospective in various domains, such as manipulation and interactions with humans, by offering competitive advantages against rigid robotic systems, including inherent compliance and variable stiffness. Despite these benefits, their theoretically infinite degrees of freedom and prominent nonlinearities pose significant challenges in developing dynamic models and guiding the robots along desired paths. Additionally, soft robots may exhibit rigid behaviors and potentially collide with their surroundings during path tracking tasks, particularly when possible contact points are unknown. In this dissertation, reduced-order models are used to describe the behaviors of three different soft robot designs, including both linear parameter varying (LPV) and augmented rigid robot (ARR) models. While the reduced-order model captures the majority of the soft robot's dynamics, modeling uncertainties notably remain. Non-repeated modeling uncertainties are addressed by categorizing them as a lumped disturbance, employing two methodologies, $H_\infty$ method and nonlinear disturbance observer (NDOB) based sliding mode control, for its rejection. For repeated disturbances, an iterative learning control (ILC) with a P-type learning function is implemented to enhance trajectory tracking efficacy. Furthermore,for non-repeated disturbances, the NDOB facilitates the contact estimation, and its results are jointly used with a switching algorithm to modify the robot trajectories. The stability proof of all controllers and corresponding simulation and experimental results are provided. For a path tracking task of a soft robot with multi-segments, a robust control strategy that combines a LPV model with an innovative improved nonlinear disturbance observer-based adaptive sliding mode control (INASMC). The control framework employs a first-order LPV model for dynamic representation, leverages an improved disturbance observer for accurate disturbance forecasting, and utilizes adaptive sliding mode control to effectively counteract uncertainties. The tracking error under the proposed controller is proven to be asymptotically stable, and the controller's effectiveness is is validated with simulation and experimental results. Ultimately, this research mitigates the inherent uncertainty in soft robot modeling, thereby enhancing their functionality in contact-intensive tasks.
ContributorsQIAO, ZHI (Author) / Zhang, Wenlong (Thesis advisor) / Marvi, Hamidreza (Committee member) / Lee, Hyunglae (Committee member) / Berman, Spring (Committee member) / Sugar, Thomas (Committee member) / Arizona State University (Publisher)
Created2023
Description
Navigation and mapping in GPS-denied environments, such as coal mines ordilapidated buildings filled with smog or particulate matter, pose a significant challenge
due to the limitations of conventional LiDAR or vision systems. Therefore there
exists a need for a navigation algorithm and mapping strategy which do not use vision
systems but are still able to explore and map the environment. The map can further
be used by first responders and cave explorers to access the environments.
This thesis presents the design of a collision-resilient Unmanned Aerial Vehicle
(UAV), XPLORER that utilizes a novel navigation algorithm for exploration and
simultaneous mapping of the environment. The real-time navigation algorithm uses
the onboard Inertial Measurement Units (IMUs) and arm bending angles for contact
estimation and employs an Explore and Exploit strategy. Additionally, the quadrotor
design is discussed, highlighting its improved stability over the previous design.
The generated map of the environment can be utilized by autonomous vehicles to
navigate the environment. The navigation algorithm is validated in multiple real-time
experiments in different scenarios consisting of concave and convex corners and circular
objects. Furthermore, the developed mapping framework can serve as an auxiliary
input for map generation along with conventional LiDAR or vision-based mapping
algorithms.
Both the navigation and mapping algorithms are designed to be modular, making
them compatible with conventional UAVs also. This research contributes to the
development of navigation and mapping techniques for GPS-denied environments,
enabling safer and more efficient exploration of challenging territories.
ContributorsPandian Saravanakumaran, Aravind Adhith (Author) / Zhang, Wenlong (Thesis advisor) / Das, Jnaneshwar (Committee member) / Berman, Spring (Committee member) / Arizona State University (Publisher)
Created2023
Description
In recent years, Artificial Intelligence (AI) (e.g., Deep Neural Networks (DNNs), Transformer) has shown great success in real-world applications due to its superior performance in various cognitive tasks. The impressive performance achieved by AI models normally accompanies the cost of enormous model size and high computational complexity, which significantly hampers their implementation on resource-limited Cyber-Physical Systems (CPS), Internet-of-Things (IoT), or Edge systems due to their tightly constrained energy, computing, size, and memory budget. Thus, the urgent demand for enhancing the \textbf{Efficiency} of DNN has drawn significant research interests across various communities.
Motivated by the aforementioned concerns, this doctoral research has been mainly focusing on Enabling Deep Learning at Edge: From Efficient and Dynamic Inference to On-Device Learning. Specifically, from the inference perspective, this dissertation begins by investigating a hardware-friendly model compression method that effectively reduces the size of AI model while simultaneously achieving improved speed on edge devices. Additionally, due to the fact that diverse resource constraints of different edge devices, this dissertation further explores dynamic inference, which allows for real-time tuning of inference model size, computation, and latency to accommodate the limitations of each edge device. Regarding efficient on-device learning, this dissertation starts by analyzing memory usage during transfer learning training. Based on this analysis, a novel framework called "Reprogramming Network'' (Rep-Net) is introduced that offers a fresh perspective on the on-device transfer learning problem. The Rep-Net enables on-device transferlearning by directly learning to reprogram the intermediate features of a pre-trained model. Lastly, this dissertation studies an efficient continual learning algorithm that facilitates learning multiple tasks without the risk of forgetting previously acquired knowledge. In practice, through the exploration of task correlation, an interesting phenomenon is observed that the intermediate features are highly correlated between tasks with the self-supervised pre-trained model. Building upon this observation, a novel approach called progressive task-correlated layer freezing is proposed to gradually freeze a subset of layers with the highest correlation ratios for each task leading to training efficiency.
ContributorsYang, Li (Author) / Fan, Deliang (Thesis advisor) / Seo, Jae-Sun (Committee member) / Zhang, Junshan (Committee member) / Cao, Yu (Committee member) / Arizona State University (Publisher)
Created2023
Description
Modern-day automobiles are becoming more connected and reliant on wireless connectivity. Thus, automotive electronics can be both a cause of and highly sensitive to electromagnetic interference (EMI), and the consequences of failure can be fatal. Technology advancements in engineering have brought several features into the automotive field but at the expense of electromagnetic compatibility issues. Automotive EMC problems are the result of the emissions from electronic assemblies inside a vehicle and the susceptibility of the electronics when exposed to external EMI sources. In both cases, automotive EMC problems can cause unintended changes in the automotive system operation. Robustness to electromagnetic interference (EMI) is one of the primary design aspects of state-of-the-art automotive ICs like System Basis Chips (SBCs) which provide a wide range of analog, power regulation and digital functions on the same die. One of the primary sources of conducted EMI on the Local Interconnect Network (LIN) driver output is an integrated switching DC-DC regulator noise coupling through the parasitic substrate capacitance of the SBC. In this dissertation an adaptive active EMI cancellation technique to cancel the switching noise of the DC-DC regulator on the LIN driver output to ensure electromagnetic compatibility (EMC) is presented. The proposed active EMI cancellation circuit synthesizes a phase synchronized cancellation pulse which is then injected onto the LIN driver output using an on-chip tunable capacitor array to cancel the switching noise injected via the substrate. The proposed EMI reduction technique can track and cancel substrate noise independent of process technology and device parasitics, input voltage, duty cycle, and loading conditions of the DC-DC switching regulator. The EMI cancellation system is designed and fabricated on a 180nm Bipolar-CMOS-DMOS (BCD) process with an integrated power stage of a DC-DC buck regulator at a switching frequency of 2MHz along with an automotive LIN driver. The EMI cancellation circuit occupies an area of 0.7 mm2, which is less than 3% of the overall area in a standard SBC and consumes 12.5 mW of power and achieves 25 dB reduction of conducted EMI in the LIN driver output’s power spectrum at the switching frequency and its harmonics.
ContributorsRay, Abhishek (Author) / Bakkaloglu, Bertan (Thesis advisor) / Garrity, Douglas (Committee member) / Kitchen, Jennifer (Committee member) / Seo, Jae-Sun (Committee member) / Arizona State University (Publisher)
Created2023
Description
While wearable soft robots have successfully addressed many inherent design limitations faced by wearable rigid robots, they possess a unique set of challenges due to their soft and compliant nature. Some of these challenges are present in the sensing, modeling, control and evaluation of wearable soft robots. Machine learning algorithms have shown promising results for sensor fusion with wearable robots, however, they require extensive data to train models for different users and experimental conditions. Modeling soft sensors and actuators require characterizing non-linearity and hysteresis, which complicates deriving an analytical model. Experimental characterization can capture the characteristics of non-linearity and hysteresis but requires developing a synthesized model for real-time control. Controllers for wearable soft robots must be robust to compensate for unknown disturbances that arise from the soft robot and its interaction with the user. Since developing dynamic models for soft robots is complex, inaccuracies that arise from the unmodeled dynamics lead to significant disturbances that the controller needs to compensate for. In addition, obtaining a physical model of the human-robot interaction is complex due to unknown human dynamics during walking. Finally, the performance of soft robots for wearable applications requires extensive experimental evaluation to analyze the benefits for the user. To address these challenges, this dissertation focuses on the sensing, modeling, control and evaluation of soft robots for wearable applications. A model-based sensor fusion algorithm is proposed to improve the estimation of human joint kinematics, with a soft flexible robot that requires compact and lightweight sensors. To overcome limitations with rigid sensors, an inflatable soft haptic sensor is developed to enable gait sensing and haptic feedback. Through experimental characterization, a mathematical model is derived to quantify the user's ground reaction forces and the delivered haptic force. Lastly, the performance of a wearable soft exosuit in assisting human users during lifting tasks is evaluated, and the benefits obtained from the soft robot assistance are analyzed.
ContributorsQuiñones Yumbla, Emiliano (Author) / Zhang, Wenlong (Thesis advisor) / Berman, Spring (Committee member) / Lee, Hyunglae (Committee member) / Marvi, Hamid (Committee member) / Sugar, Thomas (Committee member) / Arizona State University (Publisher)
Created2023
Description
Robotic technology can be broadly categorized into two main approaches based on the compliance of the robot's materials and structure: hard and soft. Hard, traditional robots, with mechanisms to transmit forces, provide high degrees of freedom (DoFs) and precise manipulation, making them commonly used in industry and academic research. The field of soft robotics, on the other hand, is a new trend from the past three decades of robotics that uses soft materials such as silicone or textiles as the body or material base instead of the rigid bodies used in traditional robots. Soft robots are typically pre-programmed with specific geometries, and perform well at tasks such as human-robot interaction, locomotion in complex environments, and adaptive reconfiguration to the environment, which reduces the cost of future programming and control. However, full soft robotic systems are often less mobile due to their actuation --pneumatics, high-voltage electricity or magnetics -- even if the robot itself is at a millimeter or centimeter scale. Rigid or hard robots, on the other hand, can often carry the weight of their own power, but with a higher burden of cost for control and sensing. A middle ground is thus sought, to combine soft robotics technologies with rigid robots, by implementing mechanism design principles with soft robots to embed functionalities or utilize soft robots as the actuator on a rigid robotic system towards an affordable robotic system design. This dissertation showcases five examples of this design principle with two main research branches: locomotion and wearable robotics. In the first research case, an example of how a miniature swimming robot can navigate through a granular environment using compliant plates is presented, compared to other robots that change their shape or use high DoF mechanisms. In the second pipeline, mechanism design is implemented using soft robotics concepts in a wearable robot. An origami-inspired, soft "exo-shell", that can change its stiffness on demand, is introduced. As a follow-up to this wearable origami-inspired robot, a geometry-based, ``near" self-locking modular brake is then presented. Finally, upon combining the origami-inspired wearable robot and brake design, a concept of a modular wearable robot is showcased for the purpose of answering a series of biomechanics questions.
ContributorsLi, Dongting (Author) / Aukes, Daniel M (Thesis advisor) / Sugar, Thomas G (Committee member) / Zhang, Wenlong (Committee member) / Arizona State University (Publisher)
Created2023