US 12062243

US patent 12062243 (US12062243B2) is titled "Distracted driving detection using a multi-task training process."

• Title: Distracted driving detection using a multi-task training process
• Assignee: Motive Technologies Inc.
• Inventors: Ali Hassan, Ijaz AKHTER, Muhammad FAISAL, Afsheen Rafaqat ALI, Ahmed Ali
• Filing Date: 2023-09-20
• Issue Date: 2024-08-13
• Abstract: The patent describes a multi-task training technique and a resulting model for detecting distracted driving. It involves inputting labeled data into a multi-task network, which includes a backbone network and multiple prediction heads. A joint loss function, derived from the outputs of these prediction heads, is minimized to adjust the backbone network's parameters. Finally, a distraction classification model is stored, comprising the parameters of the trained backbone network and at least one of the prediction heads.

Plain-Language Overview of Independent Claims:

• Independent Claim 1 (Method): This claim describes a method for detecting distracted driving. It involves feeding labeled examples (images/videos with known distraction statuses) into a specialized artificial intelligence (AI) network. This network has a core processing part (a "backbone network") that extracts key features from the examples, and several specialized output parts ( "prediction heads"). The method then adjusts the entire network by minimizing a combined "loss" value based on how well all the prediction heads perform their respective tasks. After this training, a final "distraction classification model" is saved, which includes the learned settings of the backbone network and at least one of the prediction heads, ready for deployment.

• Independent Claim 11 (Device): This claim outlines a device designed to detect distracted driving. The device comprises memory to store instructions and a processor to execute them. These instructions enable the device to perform the method described in Claim 1, specifically: inputting labeled examples into a multi-task network (with a backbone network generating feature vectors and coupled prediction heads), minimizing a joint loss from the prediction heads to adjust the backbone network's parameters, and storing a distraction classification model (consisting of the backbone network's parameters and those of at least one prediction head).

• Independent Claim 17 (Non-transitory Computer-Readable Storage Medium): This claim covers a non-transitory computer-readable storage medium (like a hard drive or flash memory) that contains software instructions. When these instructions are executed by a processor, they cause the processor to carry out the method for detecting distracted driving as described in Claim 1. This includes inputting labeled examples into a multi-task network (with a backbone network and prediction heads), minimizing a joint loss to modify the backbone network's parameters, and storing the resulting distraction classification model.

Litigation Status:

US patent 12062243 is involved in active litigation. Motive Technologies Inc. has filed a lawsuit against Samsara Inc., alleging patent infringement, theft of intellectual property, and false advertising, explicitly mentioning U.S. Patent No. 12,062,243 as one of the infringed patents.

Additionally, the patent information indicates:

• A PTAB (Patent Trial and Appeal Board) case, IPR2026-00108, has been filed and is pending.
• A U.S. case has been filed in the California Northern District Court (case number 3:24-cv-00902).

Regarding specific CAFC (U.S. Court of Appeals for the Federal Circuit) dockets for 2026, a direct entry for US12062243 was not found in the publicly available May or June 2026 scheduled case lists. While the patent is subject to ongoing litigation in district court and a PTAB review, a corresponding appeal specifically citing US12062243 on the CAFC's 2026 argument dockets was not identified in the search results. Further litigation in lower courts or PTAB proceedings may eventually lead to a CAFC appeal, but this has not yet appeared on the identified 2026 schedules.