Steven Bos - Academia.edu (original) (raw)
Papers by Steven Bos
2020 IEEE 8th Electronics System-Integration Technology Conference (ESTC)
This paper presents a method to read and write ternary (three-valued) signals on memristors to co... more This paper presents a method to read and write ternary (three-valued) signals on memristors to control a robotic actuator in real-time. The paper is a continuation of earlier work by [1] and implements a ternary memory controller for memristors in hardware. This post-binary approach with nonvolatile memory is used to program a memristor as a "trit". The paper contributes to the state-of-the-art in memristor controlled robotics by reporting an entropy gain of log2(3)= 58% information at (20 versus 14)= 43% more component cost compared to binary. This advantage (eg. less wire complexity) increases when multi-trit architectures are considered.This article demonstrates both an LTspice simulation of the circuit and implementation with source code. The memristor programmer circuit writes a state to the memristor using a pattern of 100us pulses at 3 different amplitudes. The memristor read circuit sends 500 nA pulses and converts these using two reference resistors to three logic levels using an op amp window comparator. An Arduino Mega microcontroller ADC pin converts the analog output to a digital trit. Strenuous effort was made for predictable and replicable applied memristor research in pursuit of a post-binary robotics era. The multiple-valued circuit has safety features to prevent harm to the memristance state, standardized forming of new memristors and programming flexibility by sending patterns of different pulse amounts, pulse width and pulse amplitudes.
This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be pro... more This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be processed with deep neural networks such that its fusion, the trained neural network, is a) more robust to noise, b) outperforms unimodal recognition and c) enhances unimodal recognition in absence of multimodal data. We built a framework for a complete workflow to experiment with multimodal sensor data ranging from recording (with Kinect 3D sensor), labeling, 3D signal processing, analysing and replaying. We also built three custom recognizers (automatic speech recognizer, 3D object recognizer and 3D gesture recognizer) to convert the raw sensor streams to decisions and feed this to the neural network using a late fusion strategy. We recorded 25 particpants performing 27 unique verbal and gestural interactions (intents) with objects and trained the neural network using a supervised strategy. We proved that the framework works by building a deep neural networks assisted speech recognizer th...
2020 IEEE 50th International Symposium on Multiple-Valued Logic (ISMVL), 2020
This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-vo... more This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-volatile ternary states to memristors. The Unity (C#) framework is a port of the open source Java project Memristor-Discovery and adds a closedloop ternary memory controller to enable both PC and realtime embedded ternary applications. We validate the closed-loop ternary memory controller in an embedded system case study with 16 M+SDC Tungsten dopant memristors. We measure an average switching speed of 3 Hz, worst case energy usage of 1 µW per switch, 0.03% random write error and no decay in (non-volatile) state retention after 15 minutes. We conclude with observations and open questions when working with memristors for ternary applications.
This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be pro... more This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be processed with deep neural networks such that its fusion, the trained neural network, is a) more robust to noise, b) outperforms unimodal recognition and c) enhances unimodal recognition in absence of multimodal data. We built a framework for a complete workflow to experiment with multimodal sensor data ranging from recording (with Kinect 3D sensor), labeling, 3D signal processing, analysing and replaying. We also built three custom recognizers (automatic speech recognizer, 3D object recognizer and 3D gesture recognizer) to convert the raw sensor streams to decisions and feed this to the neural network using a late fusion strategy. We recorded 25 particpants performing 27 unique verbal and gestural interactions (intents) with objects and trained the neural network using a supervised strategy. We proved that the framework works by building a deep neural networks assisted speech recognizer th...
This paper presents a method to read and write ternary (three-valued) signals on memristors to co... more This paper presents a method to read and write ternary (three-valued) signals on memristors to control a robotic actuator in real-time. The paper is a continuation of earlier work by [1] and implements a ternary memory controller for memristors in hardware. This post-binary approach with nonvolatile memory is used to program a memristor as a "trit". The paper contributes to the state-of-the-art in memristor controlled robotics by reporting an entropy gain of log2(3)= 58% information at (20 versus 14)= 43% more component cost compared to binary. This advantage (eg. less wire complexity) increases when multi-trit architectures are considered.This article demonstrates both an LTspice simulation of the circuit and implementation with source code. The memristor programmer circuit writes a state to the memristor using a pattern of 100us pulses at 3 different amplitudes. The memristor read circuit sends 500 nA pulses and converts these using two reference resistors to three logic...
Proceedings of the 53rd Hawaii International Conference on System Sciences, 2020
Highly-compliant elastic actuators have become progressively prominent over the last years for a ... more Highly-compliant elastic actuators have become progressively prominent over the last years for a variety of robotic applications. With remarkable shock tolerance, elastic actuators are appropriate for robots operating in unstructured environments. In accordance with this trend, a novel elastic actuator was recently designed by our research group for Serpens, a low-cost, open-source and highly-compliant multipurpose modular snake robot. To control the newly designed elastic actuators of Serpens, a two-feedback loops position control algorithm was proposed. The inner controller loop is implemented as a model reference adaptive controller (MRAC), while the outer control loop adopts a fuzzy proportional-integral controller (FPIC). The performance of the presented control scheme was demonstrated through simulations. However, the efficiency of the proposed controller is dependent on the initial values of the parameters of the MRAC controller as well as on the effort required for a human to manually construct fuzzy rules. An alternative solution to the problem might consist of using methods that do not assume a priori knowledge: a solution that derives its properties from a machine learning procedure. In this way, the controller would be able to automatically learn the properties of the elastic actuator to be controlled. In this work, a novel controller for the proposed elastic actuator is presented based on the use of an artificial neural network (ANN) that is trained with reinforcement learning. The newly designed control algorithm is extensively compared with the former approach. Simulation results are presented for both methods. The authors seek to achieve a fair, non-biased, risk-aware and trustworthy comparison.
Quality Innovation Prosperity
Purpose: In this paper we present Geocraft, a Geo-ICT framework meant to provide the information ... more Purpose: In this paper we present Geocraft, a Geo-ICT framework meant to provide the information needed to support the development of smart cities in an accessible and user-friendly way. We explored whether Geocraft could be an effective way to get the people of the place, especially youth, involved in geospatial issues. Methodology/Approach: Geocraft is a virtual environment in which we import real geospatial data into the gaming environment of the popular computer game Minecraft 1. In Geocraft, we can run real-time impact models to virtually simulate ánd visualise future developments and their implications, providing the user with relevant information during design processes. Geocraft is linked to Spatial Data Infrastructures (SDIs); data generated or added in Geocraft can upgrade existing databases and SDIs. In four use cases, Geocraft is used by children and high school students to address spatial planning challenges with the help of Geocraft. Findings: Geocraft has an appropriate level of abstraction to effectively represent the real world. The use of Geocraft enhances insight in geospatial relations and can raise awareness and insights in a number of geospatial issues. Geocraft can be used to collect the ideas of citizens, in this case children, and engage them in urban planning issues to raise solutions that can reckon on public support. Geocraft can engage thousands of children working on the same geospatial project in the same online Geocraft world. Spatial scenarios designed in Geocraft can be effectively translated into a feasible spatial design and be Research Limitation/implication: We present qualitative research results. In the next step, we will investigate how statistically significant the improvements in learning skills are. Originality/Value of paper: This paper presents a new digital environment facilitating citizen participation and educational processes. We use actual spatial data to transform physical reality into a parallel and playable virtual version of that reality. Herein we can simulate spatial processes and support collaboration. By doing so, we can provide unique visualizations of complex processes, raising insights across the borders of disciplines in an user-friendly way.
This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-vo... more This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-volatile ternary states to memristors. The Unity (C#) framework is a port of the open source Java project Memristor-Discovery and adds a closed-loop ternary memory controller to enable both PC and real-time embedded ternary applications. We validate the closed-loop ternary memory controller in an embedded system case study with 16 M+SDC Tungsten dopant memristors. We measure an average switching speed of 3 Hz, worst case energy usage of 1 μW per switch, 0.03% random write error and no decay in (non-volatile) state retention after 15 minutes. We conclude with observations and open questions when working with memristors for ternary applications.
Multi-Technology Positioning, 2017
Often, public safety services have to respond to emergency alerts of which little or nothing is k... more Often, public safety services have to respond to emergency alerts of which little or nothing is known besides the time and location, as is the case of Galileo’s SAR alerts. In such cases the emergency responders have to wait until arriving to the alert location before analysing the situation and defining an action plan, thus using precious time. Autonomous drones can be sent to the location of the alert to quickly provide real-time imagery of the situation to allow emergency responders to analyse and prepare for the situation before responding or during departure to the alert’s origin. Using a GNSS chip to know its position, and based on a geographical model on the area, it is possible to create and load a flight path for the drone to fly autonomously and perform a set of predefined actions, such as broadcasting live video stream, take pictures, drop a survival kit or establish bidirectional communications with the person that threw the alert. This chapter describes the drone-based public safety service in detail and how localisation information is used to support it.
2020 IEEE 8th Electronics System-Integration Technology Conference (ESTC)
This paper presents a method to read and write ternary (three-valued) signals on memristors to co... more This paper presents a method to read and write ternary (three-valued) signals on memristors to control a robotic actuator in real-time. The paper is a continuation of earlier work by [1] and implements a ternary memory controller for memristors in hardware. This post-binary approach with nonvolatile memory is used to program a memristor as a "trit". The paper contributes to the state-of-the-art in memristor controlled robotics by reporting an entropy gain of log2(3)= 58% information at (20 versus 14)= 43% more component cost compared to binary. This advantage (eg. less wire complexity) increases when multi-trit architectures are considered.This article demonstrates both an LTspice simulation of the circuit and implementation with source code. The memristor programmer circuit writes a state to the memristor using a pattern of 100us pulses at 3 different amplitudes. The memristor read circuit sends 500 nA pulses and converts these using two reference resistors to three logic levels using an op amp window comparator. An Arduino Mega microcontroller ADC pin converts the analog output to a digital trit. Strenuous effort was made for predictable and replicable applied memristor research in pursuit of a post-binary robotics era. The multiple-valued circuit has safety features to prevent harm to the memristance state, standardized forming of new memristors and programming flexibility by sending patterns of different pulse amounts, pulse width and pulse amplitudes.
This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be pro... more This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be processed with deep neural networks such that its fusion, the trained neural network, is a) more robust to noise, b) outperforms unimodal recognition and c) enhances unimodal recognition in absence of multimodal data. We built a framework for a complete workflow to experiment with multimodal sensor data ranging from recording (with Kinect 3D sensor), labeling, 3D signal processing, analysing and replaying. We also built three custom recognizers (automatic speech recognizer, 3D object recognizer and 3D gesture recognizer) to convert the raw sensor streams to decisions and feed this to the neural network using a late fusion strategy. We recorded 25 particpants performing 27 unique verbal and gestural interactions (intents) with objects and trained the neural network using a supervised strategy. We proved that the framework works by building a deep neural networks assisted speech recognizer th...
2020 IEEE 50th International Symposium on Multiple-Valued Logic (ISMVL), 2020
This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-vo... more This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-volatile ternary states to memristors. The Unity (C#) framework is a port of the open source Java project Memristor-Discovery and adds a closedloop ternary memory controller to enable both PC and realtime embedded ternary applications. We validate the closed-loop ternary memory controller in an embedded system case study with 16 M+SDC Tungsten dopant memristors. We measure an average switching speed of 3 Hz, worst case energy usage of 1 µW per switch, 0.03% random write error and no decay in (non-volatile) state retention after 15 minutes. We conclude with observations and open questions when working with memristors for ternary applications.
This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be pro... more This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be processed with deep neural networks such that its fusion, the trained neural network, is a) more robust to noise, b) outperforms unimodal recognition and c) enhances unimodal recognition in absence of multimodal data. We built a framework for a complete workflow to experiment with multimodal sensor data ranging from recording (with Kinect 3D sensor), labeling, 3D signal processing, analysing and replaying. We also built three custom recognizers (automatic speech recognizer, 3D object recognizer and 3D gesture recognizer) to convert the raw sensor streams to decisions and feed this to the neural network using a late fusion strategy. We recorded 25 particpants performing 27 unique verbal and gestural interactions (intents) with objects and trained the neural network using a supervised strategy. We proved that the framework works by building a deep neural networks assisted speech recognizer th...
This paper presents a method to read and write ternary (three-valued) signals on memristors to co... more This paper presents a method to read and write ternary (three-valued) signals on memristors to control a robotic actuator in real-time. The paper is a continuation of earlier work by [1] and implements a ternary memory controller for memristors in hardware. This post-binary approach with nonvolatile memory is used to program a memristor as a "trit". The paper contributes to the state-of-the-art in memristor controlled robotics by reporting an entropy gain of log2(3)= 58% information at (20 versus 14)= 43% more component cost compared to binary. This advantage (eg. less wire complexity) increases when multi-trit architectures are considered.This article demonstrates both an LTspice simulation of the circuit and implementation with source code. The memristor programmer circuit writes a state to the memristor using a pattern of 100us pulses at 3 different amplitudes. The memristor read circuit sends 500 nA pulses and converts these using two reference resistors to three logic...
Proceedings of the 53rd Hawaii International Conference on System Sciences, 2020
Highly-compliant elastic actuators have become progressively prominent over the last years for a ... more Highly-compliant elastic actuators have become progressively prominent over the last years for a variety of robotic applications. With remarkable shock tolerance, elastic actuators are appropriate for robots operating in unstructured environments. In accordance with this trend, a novel elastic actuator was recently designed by our research group for Serpens, a low-cost, open-source and highly-compliant multipurpose modular snake robot. To control the newly designed elastic actuators of Serpens, a two-feedback loops position control algorithm was proposed. The inner controller loop is implemented as a model reference adaptive controller (MRAC), while the outer control loop adopts a fuzzy proportional-integral controller (FPIC). The performance of the presented control scheme was demonstrated through simulations. However, the efficiency of the proposed controller is dependent on the initial values of the parameters of the MRAC controller as well as on the effort required for a human to manually construct fuzzy rules. An alternative solution to the problem might consist of using methods that do not assume a priori knowledge: a solution that derives its properties from a machine learning procedure. In this way, the controller would be able to automatically learn the properties of the elastic actuator to be controlled. In this work, a novel controller for the proposed elastic actuator is presented based on the use of an artificial neural network (ANN) that is trained with reinforcement learning. The newly designed control algorithm is extensively compared with the former approach. Simulation results are presented for both methods. The authors seek to achieve a fair, non-biased, risk-aware and trustworthy comparison.
Quality Innovation Prosperity
Purpose: In this paper we present Geocraft, a Geo-ICT framework meant to provide the information ... more Purpose: In this paper we present Geocraft, a Geo-ICT framework meant to provide the information needed to support the development of smart cities in an accessible and user-friendly way. We explored whether Geocraft could be an effective way to get the people of the place, especially youth, involved in geospatial issues. Methodology/Approach: Geocraft is a virtual environment in which we import real geospatial data into the gaming environment of the popular computer game Minecraft 1. In Geocraft, we can run real-time impact models to virtually simulate ánd visualise future developments and their implications, providing the user with relevant information during design processes. Geocraft is linked to Spatial Data Infrastructures (SDIs); data generated or added in Geocraft can upgrade existing databases and SDIs. In four use cases, Geocraft is used by children and high school students to address spatial planning challenges with the help of Geocraft. Findings: Geocraft has an appropriate level of abstraction to effectively represent the real world. The use of Geocraft enhances insight in geospatial relations and can raise awareness and insights in a number of geospatial issues. Geocraft can be used to collect the ideas of citizens, in this case children, and engage them in urban planning issues to raise solutions that can reckon on public support. Geocraft can engage thousands of children working on the same geospatial project in the same online Geocraft world. Spatial scenarios designed in Geocraft can be effectively translated into a feasible spatial design and be Research Limitation/implication: We present qualitative research results. In the next step, we will investigate how statistically significant the improvements in learning skills are. Originality/Value of paper: This paper presents a new digital environment facilitating citizen participation and educational processes. We use actual spatial data to transform physical reality into a parallel and playable virtual version of that reality. Herein we can simulate spatial processes and support collaboration. By doing so, we can provide unique visualizations of complex processes, raising insights across the borders of disciplines in an user-friendly way.
This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-vo... more This paper presents uMemristorToolbox, a novel open source framework that reads and writes non-volatile ternary states to memristors. The Unity (C#) framework is a port of the open source Java project Memristor-Discovery and adds a closed-loop ternary memory controller to enable both PC and real-time embedded ternary applications. We validate the closed-loop ternary memory controller in an embedded system case study with 16 M+SDC Tungsten dopant memristors. We measure an average switching speed of 3 Hz, worst case energy usage of 1 μW per switch, 0.03% random write error and no decay in (non-volatile) state retention after 15 minutes. We conclude with observations and open questions when working with memristors for ternary applications.
Multi-Technology Positioning, 2017
Often, public safety services have to respond to emergency alerts of which little or nothing is k... more Often, public safety services have to respond to emergency alerts of which little or nothing is known besides the time and location, as is the case of Galileo’s SAR alerts. In such cases the emergency responders have to wait until arriving to the alert location before analysing the situation and defining an action plan, thus using precious time. Autonomous drones can be sent to the location of the alert to quickly provide real-time imagery of the situation to allow emergency responders to analyse and prepare for the situation before responding or during departure to the alert’s origin. Using a GNSS chip to know its position, and based on a geographical model on the area, it is possible to create and load a flight path for the drone to fly autonomously and perform a set of predefined actions, such as broadcasting live video stream, take pictures, drop a survival kit or establish bidirectional communications with the person that threw the alert. This chapter describes the drone-based public safety service in detail and how localisation information is used to support it.