Acquisition, retention and transfer of heavy equipment operator skills through simulator training

Chung Yin So, Purdue University

Abstract

Initiatives and collaborations among heavy construction equipment manufacturing companies and training technology firms to develop and employ simulators for varied training purposes are becoming commonplace. However, human factors research on simulator training for operators of construction equipment is still sparse. For simulator training to be effective, it is necessary to understand how skills are learned using the simulator, how those skills are transferred to other tasks, devices, and real scenarios, and how well skills are retained after simulator training. ^ This research is on skill development, specifically as it applies to operator training for two specific types of heavy construction equipment: excavator and wheel loader. It aims at decomposing the complexity of equipment operation and distinguishing the skills to be acquired for each machine. It consists of five studies, three conducted with students at Purdue and two with expert operators at John Deere. ^ Study 1 investigated whether operation of a simulated hydraulic excavator is influenced by an intervening task performed between initial practice on the excavator and a subsequent retention test using a controls familiarization task (which involves just knowing the control functions). Two intervening tasks were inserted: practicing on a simulated loader, and reading an unrelated text intended to distract the participants. Performance on the simulator was compared against that of a group of participants who practiced on the simulated excavator throughout. The results showed no performance cost attributable to inserting practice on the simulated loader while learning the controls on the simulated excavator. The learning trends, however, prompted the question of whether the same results would bear true for learning a more complex perceptual-motor task. ^ Study 2 was intended to verify whether the alternating equipment sequence yields the same outcome for a more complex truck loading task that involves multiple operations. Besides the two experimental groups (control and loader groups) in Study 1, an additional group which was given practice on the two machines (but with a different practice schedule from the original loader group) was added to address the question of whether the duration of practice on an alternative machine affects performance on the previously learned machine. The number of sessions was also increased, from three to five, to examine the possible influence when participants continue to switch between the machines. Those participants who engaged in intervening practice on the simulated loader showed a smaller performance improvement on learning the truck loading task on the simulated excavator than did the control group who practiced on the simulated excavator for all five sessions. This outcome confirms that the controls familiarization tasks on both machines studied in the preliminary study may have been too simple for the full effects of switching between the machines to be evident. This finding of continued skill improvement upon return to the previously practiced machine inspires consideration of concurrent simulator-based training rather than the practice of learning to operate only one machine at a time. ^ Study 3 analyzed skill transfer using hierarchical task analysis (HTA) to investigate the degree of overlap in specific task components by studying the similarity and dissimilarity of the truck loading task performed in Study 2 on excavator and wheel loader simulators. After the modification and verification by operators of the initial HTAs, the finalized HTAs revealed that the lack of positive transfer found in performing the truck loading task alternately with the excavator and loader was likely due to the differences between loader and excavator in terms of the controls, physical constraints, and the explicit goals and subgoals of the task. In addition, comparing the number of levels of subgoals of HTAs did not evidence any level-of-difficulty differences between tasks. ^ Studies 4 and 5 investigated whether there is a cost when switching between different types of training modules within the same machine. Study 4 was conducted with experienced operators, who provided information on how the four selected tasks on the loader should be performed and classified the perceived difficulty level of each. Verbal protocol analysis was used to decompose the tasks of the four training modules on the loader simulator: 1) Simple Bucket Loading (B1), Filling a Trench (B2), Truck Loading (B3), and Fork Lifting (F). A nine phase, systematic method for deriving the HTAs from the think-aloud protocols was also developed in this study, which successfully generated the four HTAs. The findings show that 1) the HTA of the Fork Lifting module is significantly different than those of the three bucket loading tasks, and 2) although all three bucket loading tasks shared a similar mechanism, the operators ranked B1 as the easiest, followed by B2 and then B3 due to the corresponding accessibility of the dump targets, and fork lifting was ranked as the most difficult task. The results were used to justify the hypotheses for Study 5. ^ Study 5 sought to verify whether an alternating practice sequence within the same machine, i.e. training with an alternative tool (a wide fork) and returning to the original learned tool (a bucket) on a loader simulator, yields better skill transfer and retention (after a one-week interval). Four groups of undergraduate students were tested. Two groups were given two tasks involving bucket loading to practice in the first two sessions, whereas the other two groups were given a bucket loading task in the first session and the fork lifting task in the second session. The transfer and retention tasks both involved a bucket loading task that had not been performed in Sessions 1 and 2. The results showed that the groups who were assigned to practice on two tasks involving the manipulation of buckets performed better in the skill transfer test when the new task was introduced that also involved manipulation of the bucket. The results support the specificity of training principle (for which the practice conditions match the test conditions and thus facilitate retention or transfer) but not the progressive difficulty training principle (for which difficulty impedes performance in the learning stage but facilitates retention). It is suggested that, when training perceptual-motor tasks, tasks practiced during the learning phase should match the transfer task. Manipulation of task difficulty may play a role only if the tasks share task-relevant cognitive processes and mental models. ^ The overall findings of this research provide: 1) better understanding of skill development for the operation of construction equipment, and 2) evidence as to how the trainees can better utilize their time when training on a single machine and concurrently on multiple machines. The findings add to the general body of knowledge on perceptual-motor skill acquisition and to that on training in a specific domain via a specific technology. The findings are expected to generalize to heavy equipment training in related domains, such as forestry and mining, and domains requiring instrument handling skills and robotic arms, such as surgery and orbital space vessel external operations.^

Degree

Ph.D.

Advisors

Robert W. Proctor, Purdue University, Phillip S. Dunston, Purdue University.

Subject Area

Engineering, Industrial|Psychology, Cognitive

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS