notesum.ai

Published at November 9

CROPS: A Deployable Crop Management System Over All Possible State Availabilities

cs.AI

Released Date: November 9, 2024

Authors: Jing Wu¹, Zhixin Lai¹, Shengjie Liu¹, Suiyao Chen¹, Ran Tao¹, Pan Zhao¹, Chuyuan Tao¹, Yikun Cheng¹, Naira Hovakimyan¹

Aff.: ¹Institution not specified

Arxiv: http://arxiv.org/abs/2411.06034v1

Florida Case	N Input (kg/ha) $\downarrow$	Irrigation (L/m²) $\downarrow$	Yield (kg/ha) $\uparrow$	RF1 $\uparrow$	RF2 $\uparrow$	RF3 $\uparrow$	RF4 $\uparrow$
Empirical Baseline	360	394	10772	984	1417	1269	700
Policy1: Traditional Agent	200	120	10852	1425	1557	1538	1267
Policy1: LM-based Agent	122	192	11402	1464	1675	1590	1337
Policy1: CROPS Agent (Ours)	200	186	12281	1578	1782	1736	1420
Policy2: Traditional Agent	200	732	11244	813	1619	971	655
Policy2: LM-based Agent	160	510	11474	1126	1687	1252	999
Policy2: CROPS Agent (Ours)	160	762	13417	1155	1993	1281	1028
Policy3: Traditional Agent	19920	108	10865	-1.4e4	-1.4e4	1598	-3.0e4
Policy3: LM-based Agent	10000	264	13152	-6.1e3	-5.8e3	1788	-1.4e4
Policy3: CROPS Agent (Ours)	10120	144	13789	-5.9e3	-5.8e3	2020	-1.3e3
Policy4: Traditional Agent	160	102	10358	1398	1510	1524	1272
Policy4: LM-based Agent	160	36	10192	1428	1468	1555	1302
Policy4: CROPS Agent (Ours)	160	36	10161	1439	1479	1565	1313