Are socially-aware trajectory prediction models really socially-aware?. (August 2022)

• Record Type: Journal Article
• Authors: Saadatnejad, Saeed; Bahari, Mohammadhossein; Khorsandi, Pedram; Saneian, Mohammad; Moosavi-Dezfooli, Seyed-Mohsen; Alahi, Alexandre
• Is Part Of: Transportation research. Volume 141(2022)
• Publication Date: 2022-08
• Subjects: Human trajectory prediction -- Human social behavior simulation -- Adversarial attack -- Robustness; Transportation -- Periodicals; Transportation -- Technological innovations -- Periodicals; 388.011
• DOI: 10.1016/j.trc.2022.103705
• Languages: English
• ISSNs: 0968-090X

Abstract

Abstract: Our transportation field has recently witnessed an arms race of neural network-based trajectory predictors. While these predictors are at the core of many applications such as autonomous navigation or pedestrian flow simulations, their adversarial robustness has not been carefully studied. In this paper, we introduce a socially-attended attack to assess the social understanding of prediction models in terms of collision avoidance. An attack is a small yet carefully-crafted perturbations to fail predictors. Technically, we define collision as a failure mode of the output, and propose hard- and soft-attention mechanisms to guide our attack. Thanks to our attack, we shed light on the limitations of the current models in terms of their social understanding. We demonstrate the strengths of our method on the recent trajectory prediction models. Finally, we show that our attack can be employed to increase the social understanding of state-of-the-art models. The code is available at https://s-attack.github.io/ . Highlights: A novel adversarial attack to assess trajectory predictions' social understanding. Shedding light on the weaknesses of prediction models from different aspects. Improving the robustness properties of the predictors using our adversarial attack.