P050 Familiar locations and new locations: sleep’s role in the consolidation of spatial navigation information using a novel virtual Morris water maze task in older adults with mild cognitive impairment

Introduction The ability to navigate oneself in space is one of the first functional impairments in Alzheimer’s disease (AD). A 3D-computerised spatial navigation (SN) task was designed to delineate, for the first time in a sleep-dependent memory paradigm, egocentric and allocentric SN, the latter identified as one cognitive biomarker of AD. We examined group differences in SN memory and associations with sleep macroarchitecture. Methods Older adults with mild cognitive impairment (MCI, n=32) and controls (n=25) underwent overnight polysomnography and completed the SN task before and after sleep. Participants learnt the location of a target over 5 trials (familiar location; egocentric-dependent), then were instructed to find the target from a novel start location (allocentric-dependent). Memory % retention (MR) from both start locations were calculated by the XY coordinate of marked location to correct location of the target, pre- and post-sleep. Navigational strategies were coded using self-reported description of how participants’ found the target. Associations between MR with REM and SWS % duration, and AHI in REM and NREM were examined using Spearman’s correlations. Results Repeated-measures ANOVA showed Controls MR improved overnight whereas MCI performed worse (F=7.46, p=.009), with greatest differences on familiar start location MR (p=.02). Strategy as a covariate revealed a location by strategy interaction (p=01). Novel location MR was associated with REM%, rho=.448, (p=.02) in Controls, and REM-AHI, rho=.400 (p=.02) in MCI. Conclusion Behavioural and self-reported results suggest disrupted SN strategies relative to environment in MCI. Future studies should examine SN in association with sleep-wake neurophysiology and neuronal integrity.

EXPRESS: Resilient Allocentric Reorientation Following Transfer From the Inside to the Outside of an Arena

In two virtual spatial-navigation experiments, human participants were trained to find a hidden goal (an “internet connection” point) that was located adjacent to one of the right-angled corners of a cross-shaped virtual environment. The location of the goal was defined solely with respect to the geometry of the environmental structure. Training trials started from a single central start location (Experiment 1) or from multiple start locations over two, four or sixteen training trials (Experiment 2). Following training, participants were placed onto the outside of the same environment and asked to again find the connection point (which, unbeknown to participants, was removed) during a single test-trial. The results from both experiments revealed that participants spent more time searching in regions on the outside of the environment that were closest to the location where the hidden goal was position during the previous training stage. In contrast, participants spent very little time searching in regions whose visual appearance matched those regions that contained the hidden goal during training. These results reproduce the findings from previous research which supports the idea of an allocentric encoding of the shape of the environment during navigation, and further implies that this encoding is relatively resilient to manipulations that might be expected to undermine it.

Flash Characteristics and Precipitation Metrics of Western U.S. Lightning-Initiated Wildfires from 2017

This study examines 95 lightning-initiated wildfires and 1170 lightning flashes in the western United States between May and October 2017 to characterize lightning and precipitation rates and totals near the time of ignition. Eighty-nine percent of the wildfires examined were initiated by negative cloud-to-ground (CG) lightning flashes, and 66% of those fire starts were due to single stroke flashes. Average flash density at the fire locations was 1.1 fl km−2. The fire start locations were a median distance of 5.3 km away from the maximum flash and stroke densities in the 400 km2 area surrounding the fire start location. Fire start locations were observed to have a smaller 2-min precipitation rate and 24-h total rainfall than non-fire start locations. The median 2-min rainfall rate for fire-starting (FS) flash locations was 1.7 mm h−1, while the median for non-fire-starting (NFS) flash locations was 4.7 mm h−1. The median total 24-h precipitation value for FS flash locations was 2.9 mm, while NFS flash locations exhibited a median of 8.6 mm. Wilcoxon–Mann–Whitney rank sum testing revealed statistically different Z-Scores/p-values for the FS and NFS flash populations. These values were −5.578/1.21 × 10−8 and −7.176/3.58 × 10−13 for the 2-min precipitation rate and 24-h total rainfall, respectively. Additionally, 24-h and 2-min precipitation rates were statistically significantly greater for holdover versus non-holdover fire events. The median distances between the fire start location and greatest 2-min precipitation rate and greatest 24-h precipitation total were 7.4 and 10.1 km, respectively.

IMPLEMENTATION AND FIRST EVALUATION OF AN INDOOR MAPPING APPLICATION USING SMARTPHONES AND AR FRAMEWORKS

In this paper, we present the implementation of a smartphone-based indoor mobile mapping application based on an augmented reality (AR) framework and a subsequent performance evaluation in demanding indoor environments. The implementation runs on Android and iOS devices and demonstrates the great potential of smartphone-based 3D mobile mapping. The application includes several functionalities such as device tracking, coordinate, and distance measuring as well as capturing georeferenced imagery. We evaluate our prototype system by comparing measured points from the tracked device with ground control points in an indoor environment with two different campaigns. The first campaign consists of an open, one-way trajectory whereas the second campaign incorporates a loop closure. In the second campaign, the underlying AR framework successfully recognized the start location and correctly repositioned the device. Our results show that the absolute 3D accuracy of device tracking with a standard smartphone is around 1% of the travelled distance and that the local 3D accuracy reaches sub-decimetre level.

Spatial, Temporal and Electrical Characteristics of Lightning in Reported Lightning-Initiated Wildfire Events

Analysis was performed to determine whether a lightning flash could be associated with every reported lightning-initiated wildfire that grew to at least 4 km2. In total, 905 lightning-initiated wildfires within the Continental United States (CONUS) between 2012 and 2015 were analyzed. Fixed and fire radius search methods showed that 81–88% of wildfires had a corresponding lightning flash within a 14 day period prior to the report date. The two methods showed that 52–60% of lightning-initiated wildfires were reported on the same day as the closest lightning flash. The fire radius method indicated the most promising spatial results, where the median distance between the closest lightning and the wildfire start location was 0.83 km, followed by a 75th percentile of 1.6 km and a 95th percentile of 5.86 km. Ninety percent of the closest lightning flashes to wildfires were negative polarity. Maximum flash densities were less than 0.41 flashes km2 for the 24 h period at the fire start location. The majority of lightning-initiated holdover events were observed in the Western CONUS, with a peak density in north-central Idaho. A twelve day holdover event in New Mexico was also discussed, outlining the opportunities and limitations of using lightning data to characterize wildfires.

Toward Self-Driving Bicycles Using State-of-the-Art Deep Reinforcement Learning Algorithms

In this paper, we propose a controller for a bicycle using the DDPG (Deep Deterministic Policy Gradient) algorithm, which is a state-of-the-art deep reinforcement learning algorithm. We use a reward function and a deep neural network to build the controller. By using the proposed controller, a bicycle can not only be stably balanced but also travel to any specified location. We confirm that the controller with DDPG shows better performance than the other baselines such as Normalized Advantage Function (NAF) and Proximal Policy Optimization (PPO). For the performance evaluation, we implemented the proposed algorithm in various settings such as fixed and random speed, start location, and destination location.

Cost-Based Goal Recognition in Navigational Domains

Goal recognition is the problem of determining an agent's intent by observing her behaviour. Contemporary solutions for general task-planning relate the probability of a goal to the cost of reaching it. We adapt this approach to goal recognition in the strict context of path-planning. We show (1) that a simpler formula provides an identical result to current state-of-the-art in less than half the time under all but one set of conditions. Further, we prove (2) that the probability distribution based on this technique is independent of an agent's past behaviour and present a revised formula that achieves goal recognition by reference to the agent's starting point and current location only. Building on this, we demonstrate (3) that a Radius of Maximum Probability (i.e., the distance from a goal within which that goal is guaranteed to be the most probable) can be calculated from relative cost-distances between the candidate goals and a start location, without needing to calculate any actual probabilities. In this extended version of earlier work, we generalise our framework to the continuous domain and discuss our results, including the conditions under which our findings can be generalised back to goal recognition in general task-planning.

Sequential drilling and drill angulation reduce the accuracy of drill hole start location in a synthetic bone model

The accuracy of drill hole location is critical for implant placement in orthopaedic surgery. Increasing drill bit size sequentially has been suggested as a method for improving the accuracy of drill hole start location. The aim of this study was to determine whether sequential drilling or drill angulation would alter accuracy of drill hole start location. Three specialist veterinary surgeons drilled holes in synthetic bone models either directly, or with sequentially increasing drill bit sizes. Drilling was performed at 0o, 10o and 20o to perpendicular to the bone models. Three synthetic bone models were used to mimic canine cancellous and cortical bones. Sequential drilling resulted in greater inaccuracy in drill hole location when assessing all drilling angles together. There was no influence of surgeon or synthetic bone density on drilling accuracy. The combination of drill angulation and sequential drilling increased inaccuracy in drill hole start location. We conclude that sequential drilling decreased accuracy of drill hole location in the synthetic bone model when drilling was angled. Inaccuracy associated with the drill hole start location should be taken into account when performing surgery, although the magnitude of inaccuracy is low when compared with other sources of error such as angulation.