Low-latency Feedback Research Articles

Low Latency Feedback: Contrasting Training Outcomes of Immediate versus Delayed Feedback in Speedrunning Super Mario Bros.

With the rise of popularity and societal importance of video games, trends of grinding play for excessive amounts of time to improve proficiency becomes more and more common. This bears not only physical and mental health risks, but has already shown to be detrimental for skill acquisition, defeating its purpose in the first place. For the sake of health promotion, well-being, and the advancement of self-regulated learning, we specifically investigate the impact of detailed feedback on player performance. As research about comparable activities in related domains suggest inconclusive implications for when feedback in video game learning should be presented, we particularly explore the impact of delivering immediate versus delayed feedback, during or after play. Using (n = 30) participants and the well-investigated Super Mario Bros., we communicate insights in favor for deploying feedback immediately when polishing one's video game related skills such as in speedrunning.

A fast, lightweight deep learning vision pipeline for autonomous UAV landing support with added robustness

Despite massive development in aerial robotics, precise and autonomous landing in various conditions is still challenging. This process is affected by many factors, such as terrain shape, weather conditions, and the presence of obstacles. This paper describes a deep learning-accelerated image processing pipeline for accurate detection and relative pose estimation of the UAV with respect to the landing pad. Moreover, the system provides increased safety and robustness by implementing human presence detection and error estimation for both landing target detection and pose computation. Human presence and landing pad location are performed by estimating the presence probability via segmentation. This is followed by the landing pad keypoints’ location regression algorithm, which, in addition to coordinates, provides the uncertainty of presence for each defined landing pad landmark. To perform the aforementioned tasks, a set of lightweight neural network models was selected and evaluated. The resulting measurements of the system’s performance and accuracy are presented for each component individually and for the whole processing pipeline. The measurements are performed using onboard embedded UAV hardware and confirm that the method can provide accurate, low-latency feedback information for safe landing support.

Realizing a deep reinforcement learning agent for real-time quantum feedback

Realizing the full potential of quantum technologies requires precise real-time control on time scales much shorter than the coherence time. Model-free reinforcement learning promises to discover efficient feedback strategies from scratch without relying on a description of the quantum system. However, developing and training a reinforcement learning agent able to operate in real-time using feedback has been an open challenge. Here, we have implemented such an agent for a single qubit as a sub-microsecond-latency neural network on a field-programmable gate array (FPGA). We demonstrate its use to efficiently initialize a superconducting qubit and train the agent based solely on measurements. Our work is a first step towards adoption of reinforcement learning for the control of quantum devices and more generally any physical device requiring low-latency feedback.

Acoustic cloning

Cloning refers to producing copies of objects that cannot be distinguished from the original based on their behavior or response. Here, we present a general methodology to clone objects that scatter acoustic waves and demonstrate it experimentally. We acquire digital twins and bring them back to life—a simple two-step process. First, we place the scattering object in a circular receiver aperture and insonify it from the outside using simple speakers. From the recorded data, which may be reverberative, we retrieve the object's scattering Green's functions for radiation conditions using a technique called multi-dimensional deconvolution. This process recovers the temporal and spatial bandwidth and removes the scattering interactions with the boundaries of the experimentation domain, if present. In the second step, the acoustic scatterer is removed and reconstructed holographically using the acquired scattering Green's functions. The hologram scatters arbitrary incident wavefields in real-time exactly like the original object. Low-latency feedback enables reproducing all orders of wave interactions between physical scatterers and the numerical hologram. The two-step process is demonstrated by cloning several rigid scatterers in a two-dimensional acoustic waveguide. Applications range from fully realistic digital scattering models to efficient meta-material experimentation. [Work supported by SNSF grant 197182.]

Improved Short-Term Wind Power Forecasts: Low-Latency Feedback Error Correction Using Ramp Prediction and Data From Nearby Farms

Improved Short-Term Wind Power Forecasts: Low-Latency Feedback Error Correction Using Ramp Prediction and Data From Nearby Farms

An Autonomous Hand Hygiene Tracking Sensor System for Prevention of Hospital Associated Infections

Hospital associated infection (HAI) can lead to serious medical issues in healthcare systems, which is a major concern in the recent era. Hand hygiene is considered as an effective measure to prevent cross infections in hospital environments. Research efforts in the field of Internet of Things (IoT) can provide reliable edge-based solutions in hospitals to prevent HAI by adopting sensing, networking, and computing technologies. In this paper, we present a prototype for autonomous hand hygiene tracking combining different IoT technologies, which can be applied in healthcare environments to monitor hand hygiene activities and better prevent HAIs. The system can record hand-washing activities and provide prompt feedback if the hand-washing is not performed as required using wearable devices and smart dispensers. Collected data are transferred to a gateway via Bluetooth Low Energy (BLE) and Long Range (LoRa) wireless technologies for long-range network coverage. The hand-washing data are processed at the edge of the network (i.e., router and gateway) to provide low-latency feedback. The data can also be forwarded to a cloud server for storage, analysis, and visualization. The proposed hand hygiene tracking prototype has been tested and evaluated under different scenarios in a mock ward, demonstrating its applicability in hospital environments.

Biology-guided radiotherapy: redefining the role of radiotherapy in metastatic cancer.

The emerging biological understanding of metastatic cancer and proof-of-concept clinical trials suggest that debulking all gross disease holds great promise for improving patient outcomes. However, ablation of multiple targets with conventional external beam radiotherapy systems is burdensome, which limits investigation and utilization of complete metastatic ablation in the majority of patients with advanced disease. To overcome this logistical hurdle, technical innovation is necessary. Biology-guided radiotherapy (BgRT) is a new external beam radiotherapy delivery modality combining positron emission tomography-computed tomography (PET-CT) with a 6 MV linear accelerator. The key innovation is continuous response of the linear accelerator to outgoing tumor PET emissions with beamlets of radiotherapy at subsecond latency. This allows the deposited dose to track tumors in real time. Multiple new hardware and algorithmic advances further facilitate this low-latency feedback process. By transforming tumors into their own fiducials after intravenous injection of a radiotracer, BgRT has the potential to enable complete metastatic ablation in a manner efficient for a single patient and scalable to entire populations with metastatic disease. Future trends may further enhance the utility of BgRT in the clinic as this technology dovetails with other innovations in radiotherapy, including novel dose painting and fractionation schemes, radiomics, and new radiotracers.

A wireless millimetre-scale implantable neural stimulator with ultrasonically powered bidirectional communication.

Clinically approved neural stimulators are limited by battery requirements, as well as by their large size compared with the stimulation targets. Here, we describe a wireless, leadless and battery-free implantable neural stimulator that is 1.7 mm3 and that incorporates a piezoceramic transducer, an energy-storage capacitor and an integrated circuit. An ultrasonic link and a hand-held external transceiver provide the stimulator with power and bidirectional communication. The stimulation protocols were wirelessly encoded on the fly, reducing power consumption and on-chip memory, and enabling protocol complexity with a high temporal resolution and low-latency feedback. Uplink data indicating whether stimulation occurs are encoded by the stimulator through backscatter modulation and are demodulated at the external transceiver. When embedded in ex vivo porcine tissue, the integrated circuit efficiently harvested ultrasonic power, decoded downlink data for the stimulation parameters and generated current-controlled stimulation pulses. When cuff-mounted and acutely implanted onto the sciatic nerve of anaesthetized rats, the device conferred repeatable stimulation across a range of physiological responses. The miniaturized neural stimulator may facilitate closed-loop neurostimulation for therapeutic interventions.

Respiratory Motion Prediction Using Fusion-Based Multi-Rate Kalman Filtering and Real-Time Golden-Angle Radial MRI.

Magnetic resonance imaging (MRI) can provide guidance for interventions in organs affected by respiration (e.g., liver). This study aims to: 1) investigate image-based and surrogate-based motion tracking methods using real-time golden-angle radial MRI; and 2) propose and evaluate a new fusion-based respiratory motion prediction framework with multi-rate Kalman filtering. Images with different temporal footprints were reconstructed from the same golden-angle radial MRI data stream to simultaneously enable image-based and surrogate-based tracking at 10Hz. A custom software pipeline was constructed to perform online tracking and calibrate tracking error and latency using a programmable motion phantom. A fusion-based motion prediction method was developed to combine the lower tracking error of image-based tracking with the lower latency of surrogate-based tracking. The fusion-based method was evaluated in retrospective studies using in vivo real-time free-breathing liver MRI. Phantom experiments confirmed that the median online tracking error of image-based tracking was lower than surrogate-based methods, however, with higher median system latency (350 ms vs. 150 ms). In retrospective in vivo studies, 75 respiratory waveforms of target features from eight subjects were analyzed. The median root-mean-squared prediction error (RMSE) for the fusion-based method (0.97mm) was reduced (Wilcoxon signed rank test p < 0.05) compared to surrogate-based (1.18mm) and image-based (1.3mm) methods. The proposed fusion-based respiratory motion prediction framework using golden-angle radial MRI can achieve low-latency feedback with improved accuracy. Respiratory motion prediction using the fusion-based method can overcome system latency to provide accurate feedback information for MRI-guided interventions.

Digital Coherent Control of a Superconducting Qubit

High-fidelity gate operations are essential to the realization of a fault-tolerant quantum computer. In addition, the physical resources required to implement gates must scale efficiently with system size. A longstanding goal of the superconducting qubit community is the tight integration of a superconducting quantum circuit with a proximal classical cryogenic control system. Here we implement coherent control of a superconducting transmon qubit using a Single Flux Quantum (SFQ) pulse driver cofabricated on the qubit chip. The pulse driver delivers trains of quantized flux pulses to the qubit through a weak capacitive coupling; coherent rotations of the qubit state are realized when the pulse-to-pulse timing is matched to a multiple of the qubit oscillation period. We measure the fidelity of SFQ-based gates to be ~95% using interleaved randomized benchmarking. Gate fidelities are limited by quasiparticle generation in the dissipative SFQ driver. We characterize the dissipative and dispersive contributions of the quasiparticle admittance and discuss mitigation strategies to suppress quasiparticle poisoning. These results open the door to integration of large-scale superconducting qubit arrays with SFQ control elements for low-latency feedback and stabilization.

NFBLab-A Versatile Software for Neurofeedback and Brain-Computer Interface Research.

Neurofeedback (NFB) is a real-time paradigm, where subjects learn to volitionally modulate their own brain activity recorded with electroencephalographic (EEG), magnetoencephalographic (MEG) or other functional brain imaging techniques and presented to them via one of sensory modalities: visual, auditory or tactile. NFB has been proposed as an approach to treat neurological conditions and augment brain functions. Although the early NFB studies date back nearly six decades ago, there is still much debate regarding the efficiency of this approach and the ways it should be implemented. Partly, the existing controversy is due to suboptimal conditions under which the NFB training is undertaken. Therefore, new experimental tools attempting to provide optimal or close to optimal training conditions are needed to further exploration of NFB paradigms and comparison of their effects across subjects and training days. To this end, we have developed open-source NFBLab, a versatile, Python-based software for conducting NFB experiments with completely reproducible paradigms and low-latency feedback presentation. Complex experimental protocols can be configured using the GUI and saved in NFBLab's internal XML-based language that describes signal processing tracts, experimental blocks and sequences including randomization of experimental blocks. NFBLab implements interactive modules that enable individualized EEG/MEG signal processing tracts specification using spatial and temporal filters for feature selection and artifacts removal. NFBLab supports direct interfacing to MNE-Python software to facilitate source-space NFB based on individual head models and properly tailored individual inverse solvers. In addition to the standard algorithms for extraction of brain rhythms dynamics from EEG and MEG data, NFBLab implements several novel in-house signal processing algorithms that afford significant reduction in latency of feedback presentation and may potentially improve training effects. The software also supports several standard BCI paradigms. To interface with external data acquisition devices NFBLab employs Lab Streaming Layer protocol supported by the majority of EEG vendors. MEG devices are interfaced through the Fieldtrip buffer.

PIE: A Domain-Specific Language for Interactive Software Development Pipelines

Context. Software development pipelines are used for automating essential parts of software engineering processes, such as build automation and continuous integration testing. In particular, interactive pipelines, which process events in a live environment such as an IDE, require timely results for low-latency feedback, and persistence to retain low-latency feedback between restarts. Inquiry. Developing an incrementalized and persistent version of a pipeline is one way to reduce feedback latency, but requires dependency tracking, cache invalidation, and other complicated techniques. Therefore, interactivity complicates pipeline development if timeliness and persistence become responsibilities of the pipeline programmer, rather than being supported by the underlying system. Approach. We develop Pipelines for Interactive Environments (PIE), a Domain-Specific Language (DSL), API, and runtime for developing interactive software development pipelines, where ease of development is a focus. Knowledge. PIE provides a straightforward programming model that enables direct and concise expression of pipelines without boilerplate, reducing the development and maintenance effort of pipelines. Compiled pipeline programs can be embedded into interactive environments such as code editors and IDEs, enabling timely feedback at a low cost. Grounding. Compared to the state of the art, PIE reduces the code required to express an interactive pipeline by a factor of 6 in a case study on syntax-aware editors. Furthermore, we evaluate PIE in two case studies of complex interactive software development scenarios, demonstrating that PIE can handle complex interactive pipelines in a straightforward and concise way.

Bela: An embedded platform for low-latency feedback control of sound

Bela is an open-source embedded platform for audio and sensor processing. It uses a Beagle Bone Black single-board computer with a custom hard-real time audio environment based on Xenomai Linux, which is capable of submillisecond round-trip audio latencies (as low as 80 microseconds in certain configurations). The Bela hardware features stereo audio input and output, 8 channels each of 16-bit analog input and output, and 16 digital I/Os, all sampled at audio rates with nearly jitter-free alignment to the audio clock. This paper will present the hardware, software, and selected applications of Bela. Bela is suitable for creating digital musical instruments and interactive audio systems, and its low latency makes it especially well adapted for real-time feedback control over acoustic systems. It has been used in feedback control experiments with wind and string instruments and used as the basis for a study of the performer's experience of latency on percussion instruments.

Uplink HARQ for Cloud RAN via Separation of Control and Data Planes

The implementation of uplink hybrid automatic repeat request (HARQ) in a cloud-radio access network (C-RAN) architecture is constrained by the two-way latency on the fronthaul links connecting the remote radio heads (RRHs) with the baseband units (BBUs) that perform decoding. To overcome this limitation, this paper considers an architecture based on the separation of control and data planes, in which retransmission control decisions are made at the edge of the network, that is, by the RRHs or user equipment (UE), while data decoding is carried out remotely at the BBUs. This solution enables low-latency local retransmission decisions to be made at the RRHs or UE, which are not subject to the fronthaul latency constraints, while, at the same time, leveraging the decoding capability of the BBUs. A system with a BBU Hoteling system is considered first in which each RRH has a dedicated BBU in the cloud. For this system, the control-data separation leverages low-latency local feedback from an RRH to drive the HARQ process of a given UE. The throughput and the probability of error of this solution are analyzed for the three standard HARQ modes of Type-I, chase combining, and incremental redundancy over a general fading multiple-input multiple-output (MIMO) link. Then, novel user-centric low-latency feedback strategies are proposed and analyzed for the C-RAN architecture, with a single centralized BBU, based on limited “hard” or “soft” local feedback from the RRHs to the UE and on retransmission decisions taken at the UE. The analysis presented in this paper allows the optimization of the considered schemes, as well as the investigation of the impact of system parameters such as HARQ protocol type, blocklength, and number of antennas on the performance of low-latency local HARQ decisions in BBU Hoteling and C-RAN architectures.

Design space exploration of hardware platforms for interactive low latency movement sonification

Objective real-time acoustic feedback of complex movements arises in applications ranging from assisted training in sport to medical rehabilitation. This so called interactive movement sonification enables continuous feedback. Multiple studies have shown the benefit of sonification in movement relearning. Ubiquitous feedback requires a mobile and wearable hardware platform independent from any application. Therefore, in this paper a design space exploration regarding the design objectives power dissipation and latency is performed. Especially in sports or medical rehabilitation a low-latency feedback is crucial. Furthermore, low power dissipation mobile devices enable the utilization of the sonification technique within new applications. The most computational intensive task within the data-flow is real-time interactive sound synthesis. In contrast to commercial processors, customizable processors enable execution time reduction of critical sections by integration of application specific instruction set extensions. This paper aims to show the benefits from processor customization, as well as the location of consumer electronics devices within the design space.