Pathway I: Energy Efficient Human Behavior
Most occupants of commercial buildings are not aware of the impact of their individual activities and decisions on energy consumption. Thus, the first goal of this proposal is to educate building occupants of the energy consequences of their actions. We propose to educate building occupants about their energy consumption and the ways that they can make concrete behavioral changes to achieve greater energy efficiency.
Design of Behavioral Interventions to Reduce Standby Energy Use in Buildings
Research in the area of social influence has suggested that individuals interpret issues in line with the values and ideology of their social groups, leading them to adopt the group’s attitudes on these topics. This suggests that linking environmentally conscious behaviors to positive in-group identities and environmentally unconscionable behaviors to negative out-group identities could provide an effective means of influencing conservation behaviors. Keeping that in mind, in this project we aim to target an important energy-use behavior in which the wasteful use of standby power accrued by leaving powered down electronics plugged-in.
In two freshman dormitories at the University of Southern California, we posted flyers that either provided information about the negative effects of using standby power (control condition), or depicted standby energy use as the purview of graduate students and older, technologically-less-savvy individuals, while linking unplugging behavior to younger, socially conscious undergraduates (social identity condition). Two weeks later, we assessed group identity (how much individuals wanted to represent and endorse the attitudes and values of USC undergraduate and graduate students, respectively), interpretation of unplugging behavior (e.g., as socially responsible and forward thinking or a waste of time), and how often students unplugged their electronics when they were powered down.
Results: Coming soon
Publications:
Hayes, T. & Wood, W. (2014). Social influence depends on social meaning. (Manuscript in prep for submission to Journal of Personality and Social Psychology)
We developed THINC (agent Tool for Human INcentivization and Cooperation), an agent for saving energy in real-world commercial buildings (Pathway I and II). We address two critical issues, centred on users, that are essential in real-world agent deployments: (i) incentivizing users for their energy saving activities and (ii) interacting with users to reschedule key “energy-consuming” meetings in a timely fashion, while handling the uncertainty in such interactions. THINC addresses these issues by providing four new major contributions. First, THINC computes fair division of credits from energy savings. For this fair division, THINC provides novel algorithmic advances for efficient computation fair credits via the Shapley value based fair credit allocation scheme. Second, THINC includes a novel robust BM-MDP algorithm to optimally reschedule identified key meetings addressing user interaction uncertainty. Third, THINC provides an end-to-end integration within a single agent of energy efficient scheduling, rescheduling and credit allocation. We deployed THINC in the real-world as a pilot project at one of the main libraries at the University of Southern California and presented results illustrating the benefits in saving energy.
While THINC performs energy-efficient scheduling, it may perceive that shifting some carefully selected meetings can lead to significant energy savings. THINC then makes suggestions to involved users on how to best reschedule their meetings while ensuring a balance between energy savings and user comfort (hence, our multi-objective MDP). We cannot know the exact likelihood that users will comply with suggestions, and we may also be uncertain about the reward from energy-savings and user comfort (hence the model uncertainty). THINC employs a BM-MDP (Bounded-parameter Multi-objective Markov Decision Problem) in reallocating key energy-consuming meetings; it presents two new algorithms for BM-MDPs to handle the two layers of uncertainty: (i) model uncertainty regarding user interactions, i.e., uncertainty in user responses to reschedule requests and (ii) unexpected execution-time uncertainty regarding new meeting requests.
THINC is the first agent integrating (i) energy-efficient scheduling of user meeting requests while considering flexibility, (ii) rescheduling of key meetings for more energy savings, and (iii) fair credit allocations based on Shapley value to incentivize users for their energy saving activities (i.e., providing flexibility). We deployed it in the real world and performed experiments and obtained significant results.
Developing an Integrated Agent to Conserve Energy in Commercial Buildings
Results:
We deployed our integrated agent THINC as a pilot project at one of the main libraries at USC. We wanted to test the performance of THINC in this smaller building first before deploying it at a much bigger building where there are indeed hundreds of meetings per day. 45 students used THINC during the pilot deployment. Participants reported varying levels of time and location flexibility. The average time flexibility was 27.05%, and time flexibility ranged between 0.0% and 68.18%. The average location flexibility was 42.48%, and location flexibility ranged from 0.0 to 100.0%. This shows that, in practice, people are willing to provide a reasonable amount of flexibility allowing significant energy savings. As part of the pilot deployment, we identified 20 key meetings for rescheduling. THINC’s BM-MDP policy suggested different slots (i.e., a pair of time & location) every 6 hours. We achieved the average compliance rate of 45% for successfully rescheduling them with 3.6 interactions per user. This result clearly shows that BM-MDP for rescheduling identified meetings is useful rather than simply assuming users will blindly accept every suggestion. We then divide a portion of our energy savings based on the Shapley value. To test if the users of THINC perceived our credit allocation scheme to be fair, we asked the same participants to rate fairness and their willingness to participate in energy savings on a scale of 1 to 7, where 7 denotes a high rating for fairness and willingness to participate. The average fairness rating is 5.24 and the average willingness to participate rating is 6.0. Thus we can see that users of the system perceive the Shapley based allocation scheme to be highly fair.
Publications:
Jun-young Kwak, Debarun Kar, William B. Haskell, Pradeep Varakantham, and Milind Tambe, "Building THINC: User Incentivization and Meeting Rescheduling for Energy Savings," in Proceedings of the 13th International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May, 2014 (Innovative Applications Track)
Jun-young Kwak. The Power of Flexibility: Autonomous Agents That Conserve Energy in Commercial Buildings. (2014). University of Southern California.(Thesis/Dissertations).
Red, Blue, and Green: Promoting Agreement Between Liberals and Conservatives on Environmental Issues:
Conserving energy and curbing global climate change are among the most important challenges facing the world’s population in the 21st century. Although millions of people –such as those with environmentalist and liberal views – are already making a concerted effort to engage in a variety of environmentally beneficial behaviors, not everyone shares these concerns. How can conservative, non-environmentalists be persuaded to be more green? This seems especially difficult, given the wide schism between conservative and liberal views on important issues like the reality of global warming and the role of government in imposing regulations on businesses’ environmental practices. Despite these differences, our research suggests that environmental advocates might be most effective if they address, rather than combat, conservatives’ worldviews. That is, rather than undertaking the near-impossible task of trying to turn conservative, non-environmentalists into liberal “tree-huggers,” environmental activists should try to appeal to conservatives’ ideological values. Most conservatives will never become liberals, but they might engage in same environmentally-conscious behaviors as liberals if they are convinced those behaviors also serve their core conservative values. Thus, this program of research investigates how non-environmentalists might be influenced to construe green behaviors as in-line with their core ideological values (Pathway I).
Results:
Coming soon...
Publications:
Coming soon...
Acknowledgment and Disclaimer: This material is based upon work supported by the National Science Foundation under Grant No. 1231001. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.