{"id":{"repo_id":"mcmaster","oai_identifier":"oai:macsphere.mcmaster.ca:11375/33483"},"canonical_url":"https://search.ndltd.org/etd/mcmaster/oai:macsphere.mcmaster.ca:11375/33483","repository":{"repo_id":"mcmaster","name":"McMaster University","base_url":"https://macsphere.mcmaster.ca/server/oai/request"},"display":{"title":"Planning-Aware Trajectory Prediction: Routing Downstream Planning Risk into a Frozen-Perception Multimodal Predictor","abstract":"Autonomous vehicles forecast how surrounding agents will move and then plan a safe trajectory, yet the two stages are optimized for different things. Trajectory predictors are trained and ranked by displacement error, whereas a planner is served by forecasts that help it avoid collisions and act smoothly. Displacement is therefore a proxy for prediction quality rather than a measure of planning value, and the gap between them is usually left for the planner to absorb. This thesis introduces the Planning-Aware Loss, a plug-in, differentiable training signal that routes a downstream planning-risk gradient into the mode distribution of a multimodal trajectory predictor. The central question is whether such a signal reshapes the predictor or merely regularizes the planner that consumes its forecasts. To answer it cleanly, the loss is compared against a stop-gradient control that receives the identical objective with only the gradient path to the predictor severed, under a controlled, frozen-perception protocol on the nuScenes dataset spanning oracle and perceived input domains. The full gradient path separates from the control on all four conditions, with non-overlapping bootstrap 95% confidence intervals for a collision proxy at a displacement-controlled working point, and the separation holds across two prediction carriers, three differentiable risk surrogates, and both domains. Under a prediction-sensitive planner trained once and then frozen, open-loop displacement turns in favour of the full loss, consistent with a ghost-braking-like effect. The result is a controlled demonstration rather than a leaderboard claim, bounded to open-loop evaluation, frozen perception, and a collision-only risk term, and motivated throughout by safety at complex urban intersections.","abstract_html":"Autonomous vehicles forecast how surrounding agents will move and then plan a safe trajectory, yet the two stages are optimized for different things. Trajectory predictors are trained and ranked by displacement error, whereas a planner is served by forecasts that help it avoid collisions and act smoothly. Displacement is therefore a proxy for prediction quality rather than a measure of planning value, and the gap between them is usually left for the planner to absorb. This thesis introduces the Planning-Aware Loss, a plug-in, differentiable training signal that routes a downstream planning-risk gradient into the mode distribution of a multimodal trajectory predictor. The central question is whether such a signal reshapes the predictor or merely regularizes the planner that consumes its forecasts. To answer it cleanly, the loss is compared against a stop-gradient control that receives the identical objective with only the gradient path to the predictor severed, under a controlled, frozen-perception protocol on the nuScenes dataset spanning oracle and perceived input domains. The full gradient path separates from the control on all four conditions, with non-overlapping bootstrap 95% confidence intervals for a collision proxy at a displacement-controlled working point, and the separation holds across two prediction carriers, three differentiable risk surrogates, and both domains. Under a prediction-sensitive planner trained once and then frozen, open-loop displacement turns in favour of the full loss, consistent with a ghost-braking-like effect. The result is a controlled demonstration rather than a leaderboard claim, bounded to open-loop evaluation, frozen perception, and a collision-only risk term, and motivated throughout by safety at complex urban intersections.","abstract_has_math":false,"creators":["Guo, Xiyuan"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Electrical and Computer Engineering","school":null,"contributors":[],"advisors":["Emadi, Ali"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:36:28Z","subjects":["trajectory prediction","autonomous driving","motion planning","multimodal prediction","nuScenes"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.71548/2267"],"render_values":[{"text":"https://doi.org/10.71548/2267","href":"https://doi.org/10.71548/2267","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11375/33483","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/11375/33483","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://macsphere.mcmaster.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amacsphere.mcmaster.ca%3A11375%2F33483","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Emadi, Ali"]},{"key":"dc:contributor.department","label":"Department","values":["Electrical and Computer Engineering"]},{"key":"dc:creator","label":"Author","values":["Guo, Xiyuan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-08-24T16:18:59Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["trajectory prediction","autonomous driving","motion planning","multimodal prediction","nuScenes"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11375/33483","https://doi.org/10.71548/2267"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Autonomous vehicles forecast how surrounding agents will move and then plan a safe trajectory, yet the two stages are optimized for different things. Trajectory predictors are trained and ranked by displacement error, whereas a planner is served by forecasts that help it avoid collisions and act smoothly. Displacement is therefore a proxy for prediction quality rather than a measure of planning value, and the gap between them is usually left for the planner to absorb. This thesis introduces the Planning-Aware Loss, a plug-in, differentiable training signal that routes a downstream planning-risk gradient into the mode distribution of a multimodal trajectory predictor. The central question is whether such a signal reshapes the predictor or merely regularizes the planner that consumes its forecasts. To answer it cleanly, the loss is compared against a stop-gradient control that receives the identical objective with only the gradient path to the predictor severed, under a controlled, frozen-perception protocol on the nuScenes dataset spanning oracle and perceived input domains. The full gradient path separates from the control on all four conditions, with non-overlapping bootstrap 95% confidence intervals for a collision proxy at a displacement-controlled working point, and the separation holds across two prediction carriers, three differentiable risk surrogates, and both domains. Under a prediction-sensitive planner trained once and then frozen, open-loop displacement turns in favour of the full loss, consistent with a ghost-braking-like effect. The result is a controlled demonstration rather than a leaderboard claim, bounded to open-loop evaluation, frozen perception, and a collision-only risk term, and motivated throughout by safety at complex urban intersections."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Applied Science (MASc)"]},{"key":"dc:title","label":"Title","values":["Planning-Aware Trajectory Prediction: Routing Downstream Planning Risk into a Frozen-Perception Multimodal Predictor"]},{"key":"others:access_status","label":"Others Access Status","values":["embargo"]}]}],"canonical_facts":{"dc:contributor.advisor":["Emadi, Ali"],"dc:contributor.department":["Electrical and Computer Engineering"],"dc:creator":["Guo, Xiyuan"],"dc:date.accessioned":["2026-08-24T16:18:59Z"],"dc:date.issued":["2026"],"dc:description.abstract":["Autonomous vehicles forecast how surrounding agents will move and then plan a safe trajectory, yet the two stages are optimized for different things. Trajectory predictors are trained and ranked by displacement error, whereas a planner is served by forecasts that help it avoid collisions and act smoothly. Displacement is therefore a proxy for prediction quality rather than a measure of planning value, and the gap between them is usually left for the planner to absorb. This thesis introduces the Planning-Aware Loss, a plug-in, differentiable training signal that routes a downstream planning-risk gradient into the mode distribution of a multimodal trajectory predictor. The central question is whether such a signal reshapes the predictor or merely regularizes the planner that consumes its forecasts. To answer it cleanly, the loss is compared against a stop-gradient control that receives the identical objective with only the gradient path to the predictor severed, under a controlled, frozen-perception protocol on the nuScenes dataset spanning oracle and perceived input domains. The full gradient path separates from the control on all four conditions, with non-overlapping bootstrap 95% confidence intervals for a collision proxy at a displacement-controlled working point, and the separation holds across two prediction carriers, three differentiable risk surrogates, and both domains. Under a prediction-sensitive planner trained once and then frozen, open-loop displacement turns in favour of the full loss, consistent with a ghost-braking-like effect. The result is a controlled demonstration rather than a leaderboard claim, bounded to open-loop evaluation, frozen perception, and a collision-only risk term, and motivated throughout by safety at complex urban intersections."],"dc:description.degree":["Master of Applied Science (MASc)"],"dc:identifier.uri":["https://hdl.handle.net/11375/33483","https://doi.org/10.71548/2267"],"dc:language.iso":["en"],"dc:subject":["trajectory prediction","autonomous driving","motion planning","multimodal prediction","nuScenes"],"dc:title":["Planning-Aware Trajectory Prediction: Routing Downstream Planning Risk into a Frozen-Perception Multimodal Predictor"],"dc:type":["Thesis"],"others:access_status":["embargo"]},"updated_at":"2026-11-18T19:36:28Z"}