April 19th 2024:
| 8:00am - 9:00am | Registration and breakfast - pick-up your name tag | |
| 9:00am - 10:00am |
Keynote Address - The integration of service, science, and practice in human factors/ergonomics: One person’s perspective Camille Peres, NRC |
|
| 10:00am - 10:15am | Coffee Break Sponsored by KBR | |
| 10:15am - 11:15am | Session 1: Experiences (Chair - Vanessa Jones) (10 mins talk + 3 mins questions) Ian Robertson, KBR, An Evaluation Of The Potential Of Extended Reality Technologies For Verification Testing At Nasa At NASA, verification testing is the formal process of ensuring that a product conforms to requirements set by a project or program. Some verification methods, such as Demonstrations and Test, require either the end product or a mockup of the product with sufficient fidelity to stand-in for the product during the test. Traditionally, these mockups have been physical (e.g., foam-core and wood) but there is growing interest in exploring new methods for testing with these mockups. These methods include virtual reality (VR), mixed reality, and augmented reality which are collectively referred to as eXtended Reality (XR) technologies. VR has already been adopted and used by many in the aerospace industry as a tool for use in early design phases (e.g., developmental testing) and may have the most potential for use in verification tests. Benefits of using VR mockups offer cost effectiveness, ease of iteration, simulation of hazardous conditions (e.g., an egress through a hatch with smoke obscuring vision), and the ability to simulate microgravity conditions, which are challenging to do with physical mockups. However, the validity of test results obtained from VR mockup demonstrations or testing, compared to the current gold standard of physical mockups, remains uncertain. It is unlikely that there is one clean answer as there are many different types of verification outcomes and each XR technology must be evaluated on its own merits. This is not an issue during developmental testing as the design is still in flux and the total success of the design is not dependent upon the results of a developmental test. Verification tests, however, only happen once, assuming no change to the design, and the results are used to certify the product. Therefore, establishing the validity of XR mockup-based verification outcomes is essential before considering them for any use in verification tests. To address this concern, the Human Research Program has funded a project to explore and qualify how XR technologies might be used in verification demonstration and testing at NASA. Currently, we are conducting a review of the literature on the utilization of XR mockups for design activities, prototyping, and user testing. We are employing the Strengths, Weaknesses, Opportunities, and Threats (SWOT) analysis method to identify the pros, cons, and barriers to adoption of XR technologies for verification testing at NASA. Additionally, we are developing a framework to guide the deployment of XR mockups for verification tests. Building upon available evidence from the literature and subject-matter expert feedback, our goal for the framework is to provide guidelines for which forms of XR mockups are suitable for a given verification test, when only physical mockups should be employed and to highlight areas for which more evidence is needed. To further refine our framework and to contribute to the body of evidence, we are planning a lab-based experiment comparing a VR mockup to a physical twin for a set of select verification outcomes. Joy Xhindole, UHCL, Harnessing the Nexus of UX Design and Storytelling: The 'Subconscious Order' Case Study Humans are inherently drawn to stories, a phenomenon deeply rooted in our cognitive makeup. According to research published in the Journal of Cognitive Neuroscience, our brains are wired to process and retain information more effectively when presented in narrative form. This understanding underscores the power of storytelling in human communication. This excerpt sets the stage for a compelling case study exploring the symbiotic relationship between user experience (UX) design and story-telling. The case study centers around "The Subconcious Order" an award-winning creative campaign by Saudi Arabia’s premier food delivery app, HungerStation, that ingeniously maximizes the relationship between user psychology and technology to create a revolutionary product that tackles a universal human experience and tells a prolific story. By weaving together user-centricity and storytelling, this case study showcases how a creative narrative can be used to amplify the prolific nature of a product to revolutionize an industry and validate a shared experience. Essentially, HungerStation solved for the human factor and not merely the user problem. The story of "The Subconscious Order" showcases 2 important realities: 1) how UX professionals must remain proactive in their ultimate mission of improving the user’s experience by creating solutions to problems that go beyond just the product itself. Revolutionary work stems from exceeding the approach of creating a better product but rather how the product can solve for a broader problem. While the product serves as a conduit for either the problem or the solution, true innovation occurs when we venture beyond the confines of the product to identify the root issues and their solutions 2) the power of intersecting the realms of UX and storytelling at the point they cross paths –user psychology – to create something that is not just excellent but rather transcendent. Drawing on real-world examples and empirical research, this presentation will provide valuable insights into the strategic – and significant – integration of storytelling in product development through the amplification of user-centricity. Attendees will gain a deeper understanding of how storytelling can serve as a powerful tool for communicating complex ideas, product features and UX nuances to maximize impact. Charles Weeks, Rice University, Post-Power Law of Practice: Comparing Newer Models of Skill Acquisition Understanding the trajectory of skill acquisition as a function of practice has both theoretical and applied implications. Newell and Rosenbloom's (1981) conceptualization of skill acquisition as being described by a power function is so widely accepted that it is now often referred to as "the power law of practice." However, this view has been challenged by subsequent research by Heathcote et al. (2000) who highlighted several concerns with the original research surrounding the power law: the power function was fit to averaged data which may not accurately represent individual learning curves, and power functions were compared to exponential functions without adequately accounting for differences in their parameters. Their analysis found exponential functions fit better to individual-level data. This lecture introduces a novel approach using dynamical models to represent learning as a function of an internal state that evolves over time. Unlike traditional static models, these dynamical models allow for non-monotonic learning curves and explicitly incorporate feedback mechanisms, potentially offering a more nuanced understanding of how learning unfolds. Method - We fit models of skill acquisition to data on a motor learning task from a prior experiment. This experiment involved ninety-five undergraduate students performing a computerized mirror tracing task, where the impact of real-time haptic feedback on performance was assessed. The analysis focused on fitting task completion times to various models (three and four-parameter power, exponential, and two dynamical models) to identify which best represented individual and aggregate learning curves. The fits of the models were assessed using r2, MAD, and BIC, as each fit metric captures a different aspect of model appropriateness. Findings - We replicated previous findings that the power model fits aggregate learning curves best, but other models fit individual learning curves better. Traditional fit metrics (r2 and MAD) show the superiority of the exponential model over the power model and the new dynamical models over the traditional static models. BIC, which penalizes extra parameters to favor parsimonious models, paints a more complicated picture. Using this metric, the exponential and dynamical models both show superiority over the power models, but the dynamical models are not better than the exponential model. To our knowledge, this research is the first to apply our unique implementation of dynamical models to skill acquisition. By accounting for previous performance and feedback, these models make a different argument about the cognitive processes underlying skill acquisition than traditional static models that do not incorporate these concepts. By comparing a variety of models on a set of real-world data, we have begun the conversation about whether these models are the next advancement in the field. We hope that other researchers will start applying these models to new tasks to continue to assess their appropriateness. Xiaoxuan "Alicia" Cheng, Rice University, Dynamics of Trust in Automated Vehicles: Role of Expectations and Error Consistency Understanding drivers' trust dynamics is crucial for the effective integration and safe operation of driving automation systems (DASs). This study investigates the influence of drivers' expectations and error consistency on trust in DASs. Drivers' trust in DASs is influenced by their expectations regarding system capabilities. The Expectation Confirmation Theory (ECT), initially applied to post-purchase satisfaction in consumer studies, posits that meeting or exceeding expectations leads to satisfaction, while falling short leads to dissatisfaction. Applied to DAS, drivers evaluate systems’ performance against their expectations. Trust in automated vehicles (AVs) rises when performance exceeds expectations and declines otherwise. Drivers’ trust is also influenced by system failures. Previous studies highlight how different types of automation failures affect trust levels and subsequent trust repair. Yet, the impact of consistent failures on trust remains underexplored. Consistency in errors can establish patterns and predictability, fostering user trust and adaptive strategies. Understanding this impact informs the development of more effective DASs, an area currently lacking in research. The study will employ a between-subjects design, manipulating driver expectations (high or low) and error patterns (no errors, consistent errors, and inconsistent errors). A sample size of 162 is expected. After a practice drive, participants will navigate three distinct road segments, each triad presenting one of the error patterns. Takeover requests (TORs) will prompt manual intervention in case of errors. Participants' post-drive trust ratings and TOR reaction times will be the dependent measures. Additionally, participants will provide qualitative insights into how their trust was influenced by prior expectations and error consistency during post-task interviews. We are in the process of data collection, so the expected results are presented here. From a two-way ANOVA analysis, we expect the main effects of both expectations and error type: in the low-expectation group, we anticipate higher levels of trust and longer TOR times compared to the high-expectation group, because the system's performance exceeds participants' lower expectations; inconsistent errors will yield a lower trust and a shorter TOR than consistent errors do due to lower predictability, but both error conditions will have lower trust levels than the no-error condition because failures impair trust. Anticipated is an interaction between expectation and error pattern. In conditions of high expectations, we expect higher trust levels in the no-error condition compared to consistent errors, with the latter exhibiting higher trust and longer TOR times than inconsistent errors. Conversely, under low expectations, trust is expected to be highest in error-free condition. However, low expectations are likely to minimize differences in trust and TOR times between the two error conditions, as system performance aligns with these diminished expectations. The study enhances understanding of trust dynamics in DASs by investigating the impact of error consistency, highlighting the role of mental models. Practical insights suggest prioritizing consistency in error management, if not elimination, and transparent communication about system capabilities to foster trust and improve user experience with automated systems. |
|
| 11:15am - 12:15pm | Session 2: Panel (Chair - Hannah Bowman) Amrita Maguire from Dell Tasha David from Exxon Mobil Steven Sutherland from UHCL Sibbie Priestly Hensley from United Airlines Will Althoff from End to End User Research |
|
| 12:15am - 1:00pm | Lunch... | |
| 1:00pm - 2:00pm | Posters... | |
| 2:00pm - 3:00pm | Session 3: Cause and Effect (Chair - Andrea Wilson) (10 mins talk + 3 mins questions) Jeff Beno, MD Anderson Cancer Center, MD Anderson’s Integration of Human Factors in Incident Cause Analysis and Process Improvement The presentation will discuss MD Anderson’s integration of Human Factors (HF) tools to investigate causes of Human Error through our Cause Analysis program. To improve safety and sustainability of our process improvement solutions, we utilize Human Error Identification techniques in our safety incident analyses to guide our interventions. We’ll discuss how our Patient Safety and Healthcare Systems Engineering teams collaborate to use Root Cause Analyses and the Human Factors Classification Analysis System (HFACS) to identify human error root causes of our incidents. We’ll review how those systems have evolved over the past several years, and the ways we’ve improved the use and understanding of our Human Factors trends on an Institutional level. We’ll discuss some of our overall results, and how these efforts are feeding into understanding our processes and opportunities for improvement. We’ll also review how these techniques are infused into our Institutional training efforts on basic Human Error Identification techniques for us in their local areas to improve the effectiveness Lean, PDSA and DMAIC targeted solutions. Christine Petersen, Rice University, Effect of Cyclists' Communication Cues on Drivers' Perceptions of Intent During Cell Phone Conversation In the last decade, there has been an increase in the number of cyclists killed in traffic collisions. One potential cause of collisions between vehicles and cyclists is drivers’ inaccurate judgments of cyclists’ intentions. Another possible cause for collision is cell phone use, which diverts drivers' attention from the road. Therefore, we aimed to determine what cues drivers used to make quick and accurate judgments of intent and whether talking on a cell phone impacts drivers’ abilities to accurately predict cyclists’ intentions. Participants were shown video clips of a cyclist displaying a combination of cues, including arm signals (none, left, right, alternative right, or stop/slow), head movements (none, look right, or look left), and position on the road (left, right, or center of the lane). In each clip, the cyclist rode in a lane position for five seconds. In the subsequent two seconds, the cyclist made a head and/or arm movement, with the scene ending before the intended action was performed. After viewing each clip, participants were asked to describe the cues they used to make their decision. Half of the participants completed an additional Last Letter Task (LLT), which simulated a cell phone conversation. The participants’ speeds and accuracies in judging the cyclists' intentions were recorded. Two logistic regressions and a four-way mixed ANOVA were performed to identify if intention prediction accuracy and response time depended on arm signals, head movements, position on the road, and the LLT. Regardless of the LLT condition, drivers who saw a straight-arm signal were more likely to correctly predict the cyclist’s intention than drivers who saw a bent-arm signal. Participants who completed the LLT were less likely to correctly predict intentions than those who did not do the LLT. There was a significant two-way interaction between LLT and arm signal, F(2.71, 156.95) = 6, p less than .001, for response time. The LLT led to significantly longer response times than not performing the LLT, regardless of the arm signal. In both LLT conditions, drivers made significantly faster judgments when the cyclist used straight-arm signals than bent-arm signals. Results showed that only arm signals significantly impacted driver judgments; road position and head movement did not. This suggests that cyclists should avoid relying solely on head movement and road position when communicating with drivers. Results also showed that cyclists should avoid using bent-arm signals (i.e., stop/slow and right turn), as these signals lead to slower and less accurate judgments. Encouraging cyclists to use arm signals that are clearly understood by drivers could reduce confusion and potential collisions in the traffic environment. The finding that simulated cell phone conversations led to significantly slower and less accurate responses highlights the dangers of talking on the phone while driving. Drivers should avoid talking on the phone to increase the likelihood that they will predict cyclists’ intentions quickly and accurately. These findings have practical implications for improving road safety and emphasize the importance of clear communication between cyclists and drivers. Tianyi Mao, Rice University, Effects of Error Types and System Reliability on Drivers’ Trust and Decisions in Automated Flood Warning System The increasing threat of inland flooding due to changes in precipitation patterns and floodplain development demands the development of effective flood detection methods and communication strategies. Automated systems play a crucial role in detecting and communicating flood situations on roadways. However, these systems might produce errors, such as false alarms and misses, which can impact drivers’ trust and their compliance and reliance with the system. Research in various contexts have produced conflicting results regarding the influence of false alarms and misses on trust. Thus, this study focused on an automated flood-warning system designed for drivers to investigate the effects of error type and system reliability on trust and decision-making. Participants interacted with the flood-warning system across different driving scenarios and experienced varying error types (false alarms vs. misses) and system accuracy levels (60% vs. 90%). Their responses were analyzed to understand patterns in trust, perceived risk, perceived system reliability, and decision-making. Factorial analyses of variances (ANOVAs) were conducted on the dependent variables of subjective trust, perceived risk, and perceived system reliability, with error type and system reliability as between-subjects factors. The results revealed a significant main effect of system reliability on trust and perceived system reliability, indicating that more reliable systems were more trusted by drivers However, error type did not significantly impact trust or perceived system reliability. Similarly, perceived risk was not significantly influenced by error type or system reliability. Additionally, multinomial logistic regressions showed that higher system reliability increased the likelihood of drivers following the system’s recommendations. Moreover, false alarm-prone systems were associated with greater indecision among drivers compared to miss-prone systems. These results suggested that misses have a greater impact on people’s reliance, while false alarms have a greater impact on people’s compliance, aligning with previous studies’ findings. The findings of this study emphasize the importance of prioritizing reliability in flood-warning system design to enhance user experience and promote appropriate decision-making. Ultimately, the study underscores the significance of user-centered design principles in building flood-warning systems that foster trust, compliance, and reliance among drivers, thereby improving overall road safety. Additionally, this study provides implications for designing flood warning systems by striving for a balance between false alarms and miss rates at different system accuracy levels. Katherine Garcia, Rice University, Detail Matters: Drivers’ Decisions Given Flood Warnings in Simulated Driving Scenarios Risk communication is key in helping people make informed decisions in dangerous situations. Risk communication can vary in the detail or granularity contained in the messages. For example, a less granular and more abstract message may only contain information about the presence of a hazard, while a more granular or detailed message may include information about the location and severity of the hazard. Flood risk communication is a type of risk communication that warns people about a flood hazard and can be relayed through visual and auditory warnings. These warnings can also be conveyed through mobile devices like weather or navigation applications. The risk of driving through a flood depends both on the vehicle’s ground clearance, which is the distance from the ground to the bottom of the vehicle’s frame, and the depth of the flood. The current study tested a range of flood depths to investigate factors that affect drivers’ decisions given flood warnings from a mobile navigation application in driving simulator scenarios. A total of 85 participants were recruited through Rice University’s SONA system, a participant recruitment system. This study manipulated the type of flood warning (flood, no flood, flood of 2 inches, flood of 5 inches, flood of 5 inches maximum, flood of 8 inches; within-subjects) given to the participants. The dependent variable was the participants’ actions given a potentially flooded roadway, which was the binary option of continuing straight or accepting an alternate route and turning before encountering a potential flood. Results indicated that participants were more risk-avoidant when they were notified of the presence of any type of flood on their route compared to the no-flood condition. The percentage of risk-avoidant actions also increased as the flood depth increased. Participants acted similarly when given the general flood warning and the flood-of-8-inches warning, which may indicate a ceiling effect. From this study, providing flood depth information helped drivers accurately estimate the depth of the flood and their perceived risk of the situation. We suggest that designers include flood-depth information to aid drivers in their decision-making when faced with a flooded roadway. |
|
| 3:00pm - 3:15pm | Coffee Break | |
| 3:15pm - 4:15pm | Session 4: Evaluations (Chair - Ian Robertson ) (10 mins talk + 3 mins questions) Kritina Holden, Leidos at the NASA Johnson Space Center, User Interface Consistency: Characterization and Measurement Future space exploration missions will rely on the design and development of vehicles and complex systems from within NASA and through multiple external commercial partners to meet mission goals. Even with existing consistency-related agency requirements, NASA’s current approach to commercial spaceflight development is to allow providers flexibility and opportunities to innovate. This approach is resulting in significant design diversity across Artemis vehicles. Design best practices and guidelines champion interface consistency to facilitate mental model development and knowledge transfer. However, research investigating consistency-related benefits is mixed, and little is known about consistency in complex systems. Determining the level of risk that system diversity presents is difficult, as there is no established method for quantifying the degree of consistency within and across interfaces, nor is there information about the differential impacts of different types of inconsistency. Phase I of this project serves as a starting point to better understand the construct of consistency, its application, and the range of studies and methods used to measure it. This includes gaining awareness of the agency’s current practices to assess consistency and examples of consistency-related challenges faced within the agency. This information provided the groundwork to create the taxonomy used to develop a suite of proposed methods for assessing intersystem consistency. Checklist and cognitive walkthrough methods were developed for use by human factors (HF) and human-computer interaction (HCI) experts. The Intersystem Consistency Scale (ICS) was developed for interface evaluations with crew. A pilot study was completed with prototypes exhibiting aesthetics and interaction styles representative of interfaces that may be used by Artemis crews. The level of consistency (e.g., high, low) was manipulated between the prototype interfaces. HF/HCI experts evaluated the prototype and provided output using the checklist and the cognitive walkthrough methods. Astronaut-like participants performed tasks with the prototype using the think-aloud protocol and provided consistency ratings on the ICS. Preliminary results and next steps will be discussed. Annlyle Diokno, Rice University, Emergence of Paper-Digital Systems: A Usability Evaluation of Single-Page Versus Multipage Digital Instructions in a Ballot Mailing Task Access to food on campus is crucial for creating a delightful experience among the university community. Several food accessibility issues were identified at the Rice University campus: 1) users are not notified in changes on operating hours, 2) users do not know what food options are available until they arrive, 3) users may have difficulty locating the restaurant, and 4) users have to wait in long lines to place an order. To resolve these issues, we created an app that allows users to place orders for pickup from the restaurants at Rice University. The process began with a comparison of three interfaces: Rice Public Dining mobile website, DoorDash mobile app, and UberEats mobile app. Hierarchical task analyses for each of the three interfaces were created to break down the task into its basic actions. We then calculated and compared the execution times for four goals (i.e., locating the menu, locating the hours of operation, locating the map, and placing an order for pickup) using Cogulator, a software for creating CMN-GOMS models. The Cogulator results informed the development of a low-fidelity prototype app using Figma. The developed app outperformed the other three interfaces in terms of efficiency for all four goals. This study highlights the potential of app-based solutions in enhancing the campus dining experience. Yining (Elena) Zhang, Rice University, Unreliable Auditory Alerts' Effect on People's Performance of Identifying Phishing Emails As technology continues to advance, phishing attacks become more sophisticated and complex, resulting in a staggering number of victims and losses in the millions of dollars (Main & Bottorff, 2023). While current research efforts on anti-phishing measures have primarily focused on visual cues, individuals with visual impairments are left vulnerable. To address this issue, our study takes an empirical approach to evaluate the effectiveness of two audio alerts, speech alert and earcon (Nees & Liebman, 2023), on non-expert users. At the authors' institution, 142 participants were recruited through an online system. These individuals were tasked with reviewing 20 randomized emails, half of which were legitimate and half of which were phishing attempts. All the phishing emails were modified from legitimate emails to include suspicious or incorrect sender addresses (Chen et al., 2018). Additionally, auditory alerts were presented when certain email images appeared on the screen. The auditory alert system was designed to have a 90% correct rejection rate for phishing emails and a 42% false alarm rate for legitimate emails (Zhang et al., 2018). The independent variable in this study was the type of audio alert (control, audio-speech, or audio-earcon) between subjects, with participants' phishing detection performance and subjective ratings of the alerts serving as the dependent variables. All three groups were instructed to classify emails in their simulated inboxes as either phishing or legitimate. Each email was displayed for nine seconds before participants had the option to continue (2022 Trends in Email Engagement, 2022). After removing participants who failed attention checks, the final analysis included 117 individuals. A significant effect of alert type on the correct rejection rate was observed. The Tukey Post Hoc analysis revealed that the speech and earcon groups significantly outperformed the control group in correctly rejecting phishing emails. Both the speech and earcon groups showed significantly better performance than the control group, while no significant difference was found between the earcon and speech group. The participants also conveyed their level of trust in the audio alerts. Notably, there was no significant difference between the speech condition and the earcon condition. Nevertheless, the level of trust reported for both conditions fell significantly below the neutral mark of 50 on the trust scale of 100. Similarly, the speech and earcon group reported that they relied more on the email images than the audio alerts. According to this study, both groups that received audio alerts performed better at correctly identifying phishing emails compared to the group that did not receive any alerts. This indicates that the audio alerts effectively provided additional support to protect users from phishing emails, even when they were not completely reliable. Md Arafat Sikder, Lamar University, Assessing the impact of differing weights of cordless stick vacuums on discomfort experienced in various body parts during floor vacuuming. Cordless stick vacuums are easy to use and maneuver around furniture, making them a popular household cleaning option. Among cordless stick vacuum cleaners, two distinct styles exist: one with the center of mass (CoM) near the user's hand and another with the CoM near the brush. Upright vacuum cleaners are losing favor among home users, who increasingly prefer lightweight cordless stick vacuums, especially those with a center of mass located near the handle. Despite their popularity, user discomfort has emerged as a concern, particularly during extended cleaning sessions. This discomfort stems from the weight distribution and design of the vacuum cleaners, potentially leading to pain and fatigue in various body parts. This study investigates user discomfort in various body parts through qualitative questionnaires, focusing on the impact of weight and vacuuming duration of cordless stick cleaners. Twelve participants engaged in vacuuming tasks using three cordless stick vacuum cleaners weighing 5.2 lbs, 7.4 lbs, and 9.6 lbs respectively, on carpeted floors within an actual classroom setting. These sessions occurred at two durations: 5 minutes and 10 minutes. Discomfort levels were assessed across various body parts (shoulder, upper arm, lower back, forearm, wrist etc.) using a scale ranging from 0 (no discomfort) to 9 (unbearable discomfort). To analyze the data ANOVA was conducted, the results of this study suggest that both vacuum weight and usage time significantly impact discomfort levels in all body parts. As the weight and time increase, participants’ discomfort significantly increases. However, the strength of these effects varies depending on the body part. For example, wrist discomfort is extremely sensitive to both weight and time, while forearm discomfort is less impacted by both factors. Additionally, the interaction between weight and time is significant in some body parts, such as the shoulder, upper arm, lower back indicating that the effect of weight varies in different duration of the task. Therefore, the data emphasizes that prolonged vacuuming sessions lead to more significant discomfort in the wrist compared to other areas such as the lower back, shoulder, forearm, or upper arm. This emphasizes the need for significant ergonomic improvements in vacuum cleaner design aimed specifically at reducing discomfort in these crucial regions. |
|
| 4:15pm - 5:15pm | Session 5: Panel (Chair - Christy Harper) Lindsay Everett from Dell Michelle Spinelli from SLB Melissa Meingast from HPE Thomas Watkins from 3Leaf Consulting Natalia Russi-Vigoya from IBM Sangeeth Jeevan from UserTesting |
|
| 5:15pm - 5:30pm | Awards Ceremony (Chair - Andrew Muddimer) Presentations of: Best Paper Award sponsored by KBR Best Poster Award sponsored by KBR |
|
| After Party | 2 a days - local Pearland place Appertizers and Drinks | |
| Poster Session - Details |
1. Ameer Yadak, Texas Tech University, Replicating Basic Visuo-Spatial Distortion Effects in the Laparoscopic Training Environment
|
|
