Enhancing Environmental Engagement with Natural Language Interfaces for In-Vehicle Navigation Systems

Abstract

Four on-road studies were conducted in the Clifton area of Nottingham, UK, aiming to explore the relationships between driver workload and environmental engagement associated with ‘active’ and ‘passive’ navigation systems. In a between-subjects design, a total of 61 experienced drivers completed two experimental drives comprising the same three routes (with overlapping sections), staged one week apart. Drivers were provided with the navigational support of a commercially-available navigation device (‘satnav’), an informed passenger (a stranger with expert route knowledge), a collaborative passenger (an individual with whom they had a close, personal relationship) or a novel interface employing a conversational natural language ‘NAV-NLI’ (Navigation Natural Language Interface). The NAV-NLI was created by curating linguistic intercourse extracted from the earlier conditions and delivering this using a ‘Wizard-of-Oz’ technique. This term describes a research experiment in which subjects interact with a computer system that they believe to be autonomous, but which is actually being operated or partially operated by an unseen human being. The different navigational methods were notable for their varying interactivity and the preponderance of environmental landmark information within route directions. Participants experienced the same guidance on each of the two drives to explore changes in reported and observed behaviour. Results show that participants who were more active in the navigation task (collaborative passenger or NAV-NLI) demonstrated enhanced environmental engagement (landmark recognition, route-learning and survey knowledge) allowing them to reconstruct the route more accurately post-drive, compared to drivers using more passive forms of navigational support (SatNav or informed passenger). Workload measures (the Tactile Detection Task (TDT) and the National Aeronautical and Space Administration Task Load Index (NASA-TLX)) indicated no differences between conditions, although SatNav users and collaborative passenger drivers reported lower workload during their second drive. The research demonstrates clear benefits and potential for a navigation system employing two-way conversational language to deliver instructions. This could help support a long-term perspective in the development of spatial knowledge, enabling drivers to become less reliant on the technology and begin to re-establish associations between viewing an environmental feature and the related navigational manoeuvre.

Intelligent In‐Vehicle Interaction Technologies

“…[178] Therefore, studies have been done wherein the driver and the VA are engaged in a casual conversation rather than information gathering sequence of voice commands and have shown that it can be effective in preventing driver distraction. [179][180][181] Studies showed that short intermittent conversation with a VA can help increase driver alertness. [179] Furthermore, driving simulator-based studies has been performed to show that VAs should provide assertive voices to grab driver attention.…”

A scoping review on the use, processing and fusion of geographic data in virtual assistants

“…Linguistics refers to papers that explore text similarities, such as identifying place entities (Gupta, Ayyar, Singh, & Shah, 2018; Thenmozhi, Senthil Kumar, & Aravindan, 2018; Zhu, 2017) and spatial relationships (Prendergast & Szafir, 2018) in unstructured texts. Improving navigation is vital for location‐aware virtual assistants, be that in indoor contexts (Dobnik & Pulman, 2010; Subramani & Deepika, 2019) or outdoor spaces (Antrobus, Large, Burnett, & Hare, 2019; Cuayáhuitl & Dethlefs, 2011). Finally, improving the conversation and understanding the subtle emotions in the communication between humans and assistants (Portela & Granell‐Canut, 2017) is a primary research line regardless of the target application domain.…”

“…Among these 12 named virtual assistants, only four (SpaceBook—Bartie et al, 2018; Paval—Massai et al, 2019; TITERIA—Garrido et al, 2017; Trip 4 All—Signoretti et al, 2015) have to do with tourism‐related activities, followed by recommendation or decision‐making support (Loki—Ronzhin et al, 2019; CitiCafe—Atreja et al, 2018), healthcare (MIRob—Bandara et al, 2018; PeopleBot—González‐Medina et al, 2019), and education (CAGA—Cai et al, 2013; TreasureHuntBot—Klopfenstein et al, 2018a). Finally, Diana (Tsai, Chen, & Kang, 2019), which is related to disaster management, and Satnav (Antrobus et al, 2019), which explores relationships between driver workload and environmental engagement‐associated navigation systems, are categorized differently, even though they could still be seen under the recommendation‐related application domain.…”

“…Some kind of geospatial processing and spatial analysis is present in 40% (15/37) of the papers, corresponding mainly to tourism and recognition of places, in similar percentages, as determined in Section 3.3.1. Leaving aside user positioning, the types of spatial methods and geospatial services include route calculation and selection (Anelli et al, 2019; Antrobus et al, 2019; Subramani & Deepika, 2019), geoparsing and geocoding (Massai et al, 2019), reverse geocoding (Gintner et al, 2018), distance to POIs (Anelli et al, 2019; Sardella et al, 2019), searches for shortest paths, and optimal routes (Bartie et al, 2018). A few papers combine various geospatial methods.…”

Lessons from Oz

“…This work has revealed that an on-board agent with whom the driver can interact freely using conversational language can minimise cognitive demand and workload [7]. In addition, it can be an effective counter-measure to passive task-related fatigue [8,9], increase levels of trust and acceptance in autonomous vehicles [10,11] and enrich environmental engagement [12]. A particular challenge in conducting this work has been to create and deliver an 'authentic' CUI experience, not least to ensure that we do not violate users' high expectations of future technology or indeed how they expect, or would choose to interact.…”