Jianren Wang

About

Jianren Wang is a robotics and computer vision researcher whose work centers on perception systems for autonomous driving and robotic manipulation. His most influential contribution, "3D Multi-Object Tracking: A Baseline and New Evaluation Metrics," has accumulated over 480 citations, establishing a widely adopted framework that balances tracking accuracy with practical constraints like computational efficiency and system simplicity — a perspective often overlooked in prior literature. This work introduced new evaluation metrics that have since shaped how the community benchmarks 3D tracking systems. Beyond tracking, Wang has made meaningful contributions to sequential point cloud forecasting, probabilistic object detection using deep mixture density networks, and semi-supervised 3D object detection leveraging temporal graph neural networks — the latter addressing the costly challenge of large-scale data annotation. His research on motion prediction in visual object tracking further reflects a consistent theme of building lightweight yet effective perception pipelines. More recently, Wang has expanded into robotic manipulation, exploring how pre-trained visual representations can directly inform manipulation controllers, and how curiosity-driven tactile exploration enables robots to understand unknown objects. Together, his body of work demonstrates a cohesive vision: building robust, scalable, and uncertainty-aware robotic perception systems capable of operating reliably in complex real-world environments.

Research Focus

Key Achievements

Top Papers

1. 1
   3D Multi-Object Tracking: A Baseline and New Evaluation Metrics

   486 citations · 2020
2. 2
   3D Multi-Object Tracking: A Baseline and New Evaluation Metrics

   38 citations · 2019
3. 3
   Inverting the Pose Forecasting Pipeline with SPF2: Sequential Pointcloud\n Forecasting for Sequential Pose Forecasting

   26 citations · 2020
4. 4
   Deep Mixture Density Network for Probabilistic Object Detection

   17 citations · 2020
5. 5
   Semi-supervised 3D Object Detection via Temporal Graph Neural Networks

   16 citations · 2021
6. 6
   Motion Prediction in Visual Object Tracking

   8 citations · 2020
7. 7
   Manipulate by Seeing: Creating Manipulation Controllers from Pre-Trained Representations

   6 citations · 2023
8. 8
   Deep Mixture Density Network for Probabilistic Object Detection

   2 citations · 2019
9. 9
   Curiosity Driven Self-supervised Tactile Exploration of Unknown Objects

   2 citations · 2022
10. 10
    Semi-supervised 3D Object Detection via Temporal Graph Neural Networks

    2 citations · 2022

Key Collaborators

Contact & Links